Cloud platform management method and device, program product, and storage medium

Abstract

Embodiments of this application relate to the field of cloud computing, and provide a cloud platform management method and an apparatus, a program product, and a storage medium, to quickly adjust load pressure of a cloud platform. The method includes: grouping a plurality of cloud service components in a cloud platform into a plurality of communities based on call relationships between the plurality of cloud service components, where a degree of closeness of a call relationship between cloud service components included in each community is greater than or equal to a preset degree of closeness, and a degree of closeness between a cloud service component in any one of the plurality of communities and a cloud service component in another community is less than the preset degree of closeness; and if a target component in a first community of the plurality of communities meets a first preset condition, creating a second community, where the second community is the same as the first community, the second community is configured to share a part of load of the first community, the first preset condition indicates that performance of the target component is lower than preset performance, and the target component is at least one of a plurality of cloud service components in the first community.

Description

WIPO I PCT (12) International application published under the Patent Cooperation Treaty (19) International Bureau of the World Intellectual Property Organization (43) International Publication Date: 7 November 2024 (07.11.2024) (51) International Patent Classification: H04L 67 / 1008 (2022.01) (21) International Application Number: PCT / CN2024 / 071300 (22) International Application Date: 9 January 2024 (09.01.2024) (25) Application Language: Chinese (26) Publication Language: Chinese (30) Priority: 202310492451.0 202310820305.6 4 May 2023 (04.05.2023) CN 5 July 2023 (05.07.2023) CN (71) Applicant: Huawei Cloud Computing Technologies Co., Ltd. [CN / CN]; Huawei Cloud Data Center, Jiaoxinggong Road, Qianzhong Avenue, Gui'an New District, Guiyang City, Guizhou Province, China 550025 (CN) 0 lllllllllllllllllllllllllllllllll^ (10) International Publication No.: WO 2024 / 227365 Al (72) Inventors: Zhang Lin; Huawei Cloud Data Center, Jiaoxinggong Road, Qianzhong Avenue, Gui'an New District, Guiyang City, Guizhou Province, China 550025 (CN). Duan Bin; Huawei Cloud Data Center, Jiaoxinggong Road, Qianzhong Avenue, Gui'an New District, Guiyang City, Guizhou Province, China 550025 (CN). Wu Zhou; Huawei Cloud Data Center, Jiaoxinggong Road, Qianzhong Avenue, Gui'an New District, Guiyang City, Guizhou Province, China 550025 (CN) 0 (74) Agent: Beijing ZBSD Patent & Trademark Agency Ltd.; 8th Floor, Building 11, No. 31, Jiaoda East Road, Haidian District, Beijing, China 100044 (CN) 0 (81) Designated Country (unless otherwise specified, each of which requires national protection): AE, AG, AL, AM, AO, AT, AU, AZ, BA, BB, BG, BH, BN, BR, BW, BY, BZ, CA, CH, CL, CN, CO, CR, CU, CV, CZ, DE, DJ, DK, DM, DO, DZ, EC, EE, EG, ES, FI, = -===-= ==== ========(54) Title: Cloud Platform Management Method, Device, Program Product, and Storage Medium (54) Invention Title: A cloud platform management method, device, program product, and storage medium | AA Management Device SI 10. Obtain calling relationships of a plurality of cloud service components in a cloud platform S120 Divide into a plurality of communities S130 A target component in a first community meets a first preset condition S140 Create a second community AA Management Device BB Calling relationships CC Community A DD Community B EE Yes FF NO GG End HH Community A' (57) Abstract: Embodiments of the present application relate to the field of cloud computing, and provide a cloud platform management method and device, a program product, and a storage medium, capable of quickly adjusting the load pressure of a cloud platform. The method comprises: dividing a plurality of cloud service components into a plurality of communities according to calling relationships among the plurality of cloud service components in thecloud platform, wherein the degree of closeness of the calling relationship among the cloud service components contained in each community is greater than or equal to a preset degree of closeness, and the [See continued page] WO 2024 / 227365 Al IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIitter GB, GD, GE, GH, GM, GT, HN, HR, HU, ID, IL, IN, IQ, IR, IS, IT, JM, JO, JP, KE, KG, KH, KN, KP, KR, KW, KZ, LA, LC, LK, LR, LS, LU, LY MA, MD, MG, MK, MN, MU, MW, MX, MY MZ, NA, NG, NI, NO, NZ, OM, PA, PE, PG, PH, PL, PT, QA, RO, RS, RU, RW, SA, SC, SD, SE, SG, SK, SL, ST, SY SY TH, TJ, TM, TN, TR, TT, TZ, UA, UG, US, UZ, VC, VN, WS, ZA, ZM, ZWO (84) Designated countries (unless otherwise specified, each of the available regional protections is required): ARIPO (BW, CV, GH, GM, KE, LR, LS, MW, MZ, NA, RW, SC, SD, SL, ST, SZ, TZ, UG, ZM, ZW), Eurasia (AM, AZ, BY, KG, KZ, RU, TJ, TM), Europe (AL, AT, BE, BG, CH, CY, CZ, DE, DK, EE, ES, FI, FR, GB, GR, HR, HU, IE, IS, IT, LT, LU, LV, MC, ME, MK, MT, NL, NO, PL, PT, RO, RS, SE, SI, SK, SM, TR), OAPI (BF, BJ, CF, CG, CI, CM, GA, GN, GQ, GW, KM, ML, MR, NE, SN, TD, TG). This international publication:- Including the international search report (Article 21 of the Treaty (3)_______________________________________________________________ degree of closeness between the cloud service components in any of the plurality of communities and the cloud service components in other communities is less than the preset degree of closeness; and if a target component in a first community among the plurality of communities meets a first preset condition, creating a second community, where the second community is the same as the first community, and the second community is used for sharing part of a load of the first community, the first preset condition is used for indicating that the performance of the target component is lower than preset performance, and the target component is at least one component of the plurality of cloud service components in the first community. (57) Abstract: This application provides a cloud platform management method, apparatus, program product, and storage medium, relating to the field of cloud computing, capable of quickly adjusting the load pressure of the cloud platform. The method includes: dividing multiple cloud service components into multiple communities based on the calling relationships between them; the closeness of the calling relationships between the cloud service components contained in each community is greater than or equal to a preset closeness, and the closeness of the calling relationships between cloud service components in any one community and cloud service components in other communities is less than...The preset tightness is set; if the target component in the first community of multiple communities meets the first preset condition, a second community is created. The second community is the same as the first community and is used to share part of the load of the first community. The first preset condition is used to indicate that the performance of the target component is lower than the preset performance. The target component is at least one of the multiple cloud service components in the first community. WO 2024 / 227365 PCT / CN2024 / 071300 5 10 15 20 25 30 35 40 45 Specification A management method, apparatus, program product and storage medium for a cloud platform This application claims priority to Chinese patent applications filed on May 4, 2023, with application number 202310492451.0 and application title "A service performance analysis method and system", and on July 5, 2023, with application number 202310820305.6 and application title "A management method, apparatus, program product and storage medium for a cloud platform", the entire contents of which are incorporated herein by reference. Technical Field The embodiments of this application relate to the field of cloud computing, and more particularly to a management method, apparatus, program product and storage medium for a cloud platform. Background Technology: Cloud service is an on-demand, scalable model for providing internet-based services, enabling users to access required services on demand via the network. Cloud service implies that computing power can also be traded as a commodity through the internet. Multiple cloud services form a cloud platform, which provides powerful computing services to users. Each cloud service may consist of multiple cloud service components, which have dependencies and / or associations. With the increasing application scenarios of cloud services, customers are paying more and more attention to the processing capacity of cloud platforms or cloud services under heavy loads. A common method for handling the stress on cloud service components within a cloud platform includes: operations and maintenance personnel monitoring the out-degree and in-degree of each cloud service component and the data transmission frequency of the links between these components, and determining, based on experience, whether there are hotspots or hot links on the cloud platform. When hotspots or hot links exist, operations and maintenance personnel, based on experience, scale up the capacity of these hotspots and hot links. However, during the expansion of hotspots or hot links, operations and maintenance personnel can only adjust the load of cloud service components one by one, and cannot quickly adjust the load pressure of the cloud platform, resulting in limited cloud platform performance. This application provides a cloud platform management method, apparatus, program product, and storage medium that can quickly adjust the load pressure of the cloud platform. To achieve the above objective, this application adopts the following technical solution: Firstly, this application provides a cloud platform management method, which includes: [The method is described in the original text, but the translation is incomplete and requires further context.]The call relationships between the components are defined, and the multiple cloud service components are divided into multiple communities. Each community includes at least two cloud service components from the multiple cloud service components. The closeness of the call relationships between the cloud service components contained in each community is greater than or equal to a preset closeness. The closeness between the cloud service components in any one of the multiple communities and the cloud service components in other communities is less than the preset closeness. If the target component in the first community meets a first preset condition, a second community is created. The second community is the same as the first community and is used to share part of the load of the first community. The first preset condition is used to indicate that the performance of the target component is lower than a preset performance. The first community is any one of the multiple communities. The target component is at least one of the multiple cloud service components in the first community. Based on this, this application provides a cloud platform management method. This method divides multiple cloud service components into multiple communities, including a first community, according to the calling relationships between them. The calling relationships between cloud service components within a community are relatively close, while the calling relationships between communities are relatively close. When a first community meets preset conditions, a second community identical to the first community is created (e.g., cloning the first community to obtain a second community), so that the second community shares some of the load of the first community. Therefore, when the cloud platform is under heavy load, by cloning the entire community composed of multiple cloud service components, the high degree of calling relationships between the cloud service components within the entire community allows for load balancing between the first and second communities, which can quickly adjust the load pressure of the cloud platform and improve its performance. In one possible implementation, the calling relationships include one or more of the following: calling information of multiple cloud service components in the cloud platform, the calling frequency of the link between any two cloud service components in the cloud platform, and the out-degree and in-degree of any one of the cloud service components in the cloud platform. In one possible implementation, the aforementioned call information includes: when the first component is the main calling component, the identifier of the cloud service component called by the first component; and when the first component is the called component, the identifier of the cloud service component that calls the first component; the first component is any one of the multiple cloud service components in the aforementioned cloud platform. In another possible implementation, based on the call relationship between the multiple cloud service components in the cloud platform, the multiple cloud service components are divided into multiple communities, including: dividing the multiple cloud service components into multiple communities based on a community discovery algorithm and the call relationship.In one possible implementation, the target component includes: a cloud service component among multiple cloud service components in the first community that meets a second preset condition, or at least one cloud service component specified by the user among multiple cloud service components in the first community. In another possible implementation, the second preset condition includes: the top X cloud service components with the highest popularity values ​​among multiple cloud service components in the first community; the popularity value is the sum of the out-degree and in-degree of the cloud service component, where X is an integer greater than 0. In another possible implementation, creating a second community if the target component in the first community meets the first preset condition includes: creating a second community if the first target feature of the target component meets the first preset condition; the first target feature is at least one feature specified by the user among multiple features of the target component. In another possible implementation, the first target feature includes: at least one of processor CPU utilization, memory utilization, and I / O throughput. In one possible implementation, when the first target feature includes CPU utilization, the first preset condition includes: CPU utilization is greater than a preset CPU utilization; when the first target feature includes memory utilization, the first preset condition includes: memory utilization is greater than a preset memory utilization; when the first target feature includes I / O throughput, the first preset condition includes: I / O throughput is greater than a preset I / O throughput. In another possible implementation, creating the second community includes: obtaining the modularity of the first community with each of the other communities (excluding the first community), where modularity is a value describing the tightness of the call relationship between communities; determining whether a target community exists among the other communities based on the modularity; the target community is the community among the other communities whose modularity with the first community is greater than a threshold; and creating the second community when the target community does not exist among the other communities. In one possible implementation, when the target community exists among the other communities, a second community combination is created. This second community combination is identical to the first community combination, which includes the first community and the target community. The second community combination includes a second community and a third community, the third community being identical to the target community. The second community combination is used to share some of the load of the first community combination. This embodiment of the application determines whether there exists a target community among multiple communities whose modularity with the first community exceeds a threshold. When the target community exists (i.e., when there is a community with a high degree of call relationship with the first community), the first community and the target community are treated as a single community (i.e., the first community combination), and a second community combination, identical to the first community combination, including the second and third communities, is created. This second community combination is used to share the load of the first community combination, thereby solving the problem of excessive load on the target community caused by only expanding the first community, and further improving the effectiveness of reducing the load on the cloud platform.In one possible implementation, the method further includes: if the target component does not meet the first preset condition, then determining hotspots and hot links based on the call relationships between cloud service components in the first community; the hotspot is a cloud service component with a high call frequency among multiple cloud service components in the first community, and the hot link is a link with a high call frequency among multiple cloud service components; processing the hotspot and hot link according to a preset reduction strategy; the preset reduction strategy is used to ensure that the processed hotspot and hot link meet their respective current load requirements. In the above embodiment, when the target component in the first community does not meet the first preset condition; the hotspot and hot link in the first community are determined; then, based on the community where the hotspot is located and the third preset condition satisfied by the second target characteristic of the hotspot, a target hotspot reduction strategy is determined, and the hotspot is processed based on the target hotspot reduction strategy, thereby reducing the load pressure on the hotspot. Similarly, the management device determines the hot link reduction strategy corresponding to the fourth preset condition satisfied by the third target characteristic of the hot link as the target hot link reduction strategy, and processes the hot link according to the target hot link reduction strategy to reduce the load pressure of the processed hot link, thereby solving the problem of cloud platform downtime caused by the presence of at least one cloud service component and link with high load in the first community. In one possible implementation, the hot spots are the top N cloud service components with the highest popularity value among the multiple cloud service components included in the first community; where the popularity value is the sum of the out-degree and in-degree of the cloud service component, and N is an integer greater than 0; the hot links are the top m links with the highest call frequency among the multiple links between the multiple cloud service components included in the first community; m is an integer greater than 0. In one possible implementation, the above-mentioned hotspot processing according to a preset strategy includes: determining a target hotspot reduction strategy from a first target correspondence based on the identifier of the community where the hotspot is located and the second target feature of the hotspot; the first target correspondence includes the correspondence between the identifiers of multiple communities, multiple third preset conditions, and multiple hotspot reduction strategies; the second target feature is a feature pre-specified by the user among multiple features of the hotspot; the third preset condition is used to determine whether to expand the hotspot; and processing the hotspot according to the target hotspot reduction strategy. In another possible implementation, the above-mentioned hotspot reduction strategy includes: reset expansion, and / or horizontal expansion. In one possible implementation, the above-mentioned processing of hot links according to a preset strategy includes: determining a target hot link reduction strategy from a second target correspondence based on a third target feature of the hot link; the second target correspondence includes correspondences between multiple fourth preset conditions and multiple hot link reduction strategies; the third target feature is a feature pre-specified by the user among multiple features of the hot link.Hot links are processed according to a target hot link reduction strategy. In one possible implementation, the hot link reduction strategy includes: horizontal expansion and / or vertical expansion. In another possible implementation, processing hotspots and hot links according to a preset strategy includes: determining a target type template from multiple preset type templates based on the distribution characteristics of the hotspots and hot links; the distribution characteristics are used to characterize the spatial location information of the hotspots and hot links; the target type template is the type template with the highest matching degree with the distribution characteristics among the multiple preset type templates; determining the target reduction strategy corresponding to the target type template from the correspondence between multiple type templates and multiple reduction strategies; and processing hotspots and hot links according to the target reduction strategy. In the above embodiment, when the target component in the first community does not meet the first preset condition, hotspots and hot links in the first community are determined. Then, the management device determines the target type template from multiple preset type templates according to the distribution characteristics of the hotspots and hot links, and processes the hotspots and hot links according to the target reduction strategy corresponding to the target type template, so as to reduce the load pressure of the processed hotspots and hot links, thereby solving the problem of cloud platform downtime caused by the presence of at least one cloud service component and link with a large load in the first community. Secondly, this application provides a management device comprising: a partitioning module and a creation module; the partitioning module is configured to partition the multiple cloud service components into multiple communities based on the calling relationships between multiple cloud service components in a cloud platform; wherein each community includes at least two cloud service components from the multiple cloud service components, the closeness of the calling relationships between the cloud service components contained in each community is greater than or equal to a preset closeness, and the closeness between a cloud service component in any one of the multiple communities and a cloud service component in other communities is less than the preset closeness; the creation module is configured to create a second community if a target component in a first community of the multiple communities meets a first preset condition, the second community being the same as the first community, and the second community being used to share part of the load of the first community; the first preset condition is used to indicate that the performance of the target component is lower than a preset performance, the first community being any one of the multiple communities; the target component being at least one of the multiple cloud service components in the first community. In one possible implementation, the aforementioned call relationship includes one or more of the following: call information of multiple cloud service components in the cloud platform, call frequency of the link between any two cloud service components in the cloud platform, and out-degree and in-degree of any cloud service component in the cloud platform; wherein, the call information includes: when the first component is the main calling component, the identifier of the cloud service component called by the first component; and when the first component is the called component, the identifier of the cloud service component calling the first component.The cloud service component is identified; the first component is any one of multiple cloud service components in the cloud platform. In one possible implementation, the partitioning module is used to partition multiple cloud service components into multiple communities based on a community discovery algorithm and call relationships. In one possible implementation, the target component includes: a cloud service component in the first community that meets a second preset condition, or at least one cloud service component specified by the user in the first community. In one possible implementation, the second preset condition includes: the top X cloud service components with the highest popularity values ​​in the first community; the popularity value is the sum of the out-degree and in-degree of the cloud service component, where X is an integer greater than 0. In one possible implementation, the creation module is used to create a second community if the first target feature of the target component meets the first preset condition, where the first target feature is at least one feature specified by the user among multiple features of the target component. In one possible implementation, the first target feature includes at least one of processor CPU utilization, memory utilization, and I / O throughput. In one possible implementation, when the first target feature includes CPU utilization, the first preset condition includes: CPU utilization is greater than a preset CPU utilization; when the first target feature includes memory utilization, the first preset condition includes: memory utilization is greater than a preset memory utilization; when the first target feature includes I / O throughput, the first preset condition includes: I / O throughput is greater than a preset I / O throughput. In one possible implementation, the management device further includes: an acquisition module and a determination module; the acquisition module is used to acquire the module degree of the first community and each of the other communities in the plurality of communities other than the first community, where the module degree is a value used to describe the tightness of the calling relationship between communities; the determination module is used to determine whether a target community exists in the other communities based on the module degree; the target community is a community among the other communities whose module degree with the first community is greater than a threshold; the creation module is used to create a second community when the target community does not exist in the other communities. In another possible implementation, the creation module is used to create a second community combination when the target community exists in the other communities; the second community combination is the same as the first community combination including the first community and the target community, the second community combination includes the second community and a third community, the third community being the same as the target community, and the second community combination is used to share part of the load of the first community combination. In one possible implementation, the aforementioned determining module is used to, if the target component does not meet the first preset condition, then according to the first community...The call relationships between cloud service components determine hotspots and hot links; hotspots are cloud service components with high call frequency among multiple cloud service components in the first community, and hot links are links with high call frequency among multiple cloud service components; the aforementioned determining module is also used to process hotspots and hot links according to a preset reduction strategy; the preset reduction strategy is used to ensure that the processed hotspots and hot links meet their respective current load requirements. In one possible implementation, the aforementioned hotspots are the top N cloud service components with the highest popularity values ​​among the multiple cloud service components included in the first community; where the popularity value is the sum of the out-degree and in-degree of the cloud service component, and N is an integer greater than 0; hot links are the top m links with the highest call frequency among the multiple links among the multiple cloud service components included in the first community; m is an integer greater than 0. In one possible implementation, the determining module is used to determine a target hotspot reduction strategy from a first target correspondence based on the identifier of the community where the hotspot is located and the second target feature of the hotspot; the first target correspondence includes the identifiers of multiple communities, the correspondence between multiple third preset conditions and multiple hotspot reduction strategies; the second target feature is a feature pre-specified by the user among multiple features of the hotspot; the third preset conditions are used to determine whether to expand the hotspot; the determining module is also used to process the hotspot according to the target hotspot reduction strategy. In one possible implementation, the hotspot reduction strategy includes: reset expansion, and / or horizontal expansion. In one possible implementation, the determining module is used to determine a target hotspot link reduction strategy from a second target correspondence based on the third target feature of the hotspot link; the second target correspondence includes the correspondence between multiple fourth preset conditions and multiple hotspot link reduction strategies; the third target feature is a feature pre-specified by the user among multiple features of the hotspot link; the determining module is also used to process the hotspot link according to the target hotspot link reduction strategy. In one possible implementation, the hotspot link reduction strategy includes: horizontal expansion, and / or vertical expansion. In one possible implementation, the determining module is used to determine a target type template from multiple preset type templates based on the distribution characteristics of hotspots and hot links; the distribution characteristics are used to characterize the spatial location information of hotspots and hot links; the target type template is the type template with the highest matching degree with the distribution characteristics among the multiple preset type templates; the determining module is further used to determine the target reduction strategy corresponding to the target type template from the correspondence between multiple type templates and multiple reduction strategies; and process hotspots and hot links according to the target reduction strategy. In a third aspect, this application provides a computing device cluster, including at least one computing device, each computing device including a processor and a memory; the processor of the at least one computing device is used to execute instructions stored in the memory of the at least one computing device, so that the computing device cluster performs the method described in the first aspect and any one of its possible implementations.Fourthly, this application provides a computer-readable storage medium storing computer instructions thereon, which, when executed on a computing device, cause the computing device to perform the method described in any one of the first aspects and their possible implementations. Fifthly, this application provides a computer program product containing instructions, which, when executed on a computer, causes the computer to perform the method described in any one of the first aspects and their possible implementations. It should be understood that the beneficial effects achieved by the technical solutions of the second to fifth aspects of this application and their corresponding possible implementations can be found in the technical effects of the first aspect and its corresponding possible implementations described above, and will not be repeated here. Figure 1 is a schematic diagram of a cloud platform system provided in an embodiment of this application; Figure 2 is a schematic flowchart of a cloud platform management method provided in an embodiment of this application; Figure 3 is a schematic diagram of a call relationship graph provided in an embodiment of this application; Figure 4 is a schematic diagram of a call relationship graph provided in an embodiment of this application; Figure 5 is a schematic flowchart of a cloud platform management method provided in an embodiment of this application; Figure 6 is a schematic flowchart of a cloud platform management method provided in an embodiment of this application; Figure 7 is a schematic diagram of determining hotspots and hot links provided in an embodiment of this application; Figure 8 is a schematic flowchart of a cloud platform management method provided in an embodiment of this application; Figure 9 is a schematic flowchart of a cloud platform management method provided in an embodiment of this application; Figure 10 is a schematic diagram of a type template provided in an embodiment of this application; Figure 11 is a schematic diagram of the structure of a management device provided in an embodiment of this application; Figure 12 is a schematic diagram of a computing device cluster provided in an embodiment of this application. Figure 13 is a schematic diagram of a network connection provided in an embodiment of this application. Detailed Description: The term "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The terms "first" and "second," etc., in the specification and claims of this application are used to distinguish different objects, not to describe a specific order of objects. For example, "first community" and "second community," etc., are used to distinguish different transmission beams, not to describe a specific order of communities. In the embodiments of this application, the words "exemplary" or "for example," etc., are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of "exemplary" or "for example," etc., is intended to present related concepts in a specific manner.In the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more. For example, multiple processing units refer to two or more processing units; multiple systems refer to two or more systems. First, some concepts involved in the cloud platform management method, apparatus, program product, and storage medium provided in the embodiments of this application will be explained. Cloud service component: also known as a cloud component or component, a cloud service includes multiple cloud service components. These multiple cloud service components provide all services in the cloud service; that is, each cloud service component provides a sub-service to the cloud service. For example, suppose cloud service A is used to provide data storage services. Cloud service A includes a data caching component and a data persistence component. The data caching component provides a service for caching data to be persisted, and the data persistence component provides a service for persisting the cached data. In this case, the services provided by the data caching component and the services provided by the data persistence component together constitute cloud service A. When the cloud service component is a MySQL component, the cloud service component is used to provide MySQL database services. Out-degree and In-degree: In a directed graph, arrows have direction, pointing from one vertex to another. The number of arrows pointing to each vertex is its in-degree. The number of arrows pointing out of that vertex is its out-degree. In other words, the in-degree of a cloud service component is the number of times data is sent to that component per unit of time, or the number of times that component is called; the out-degree of a cloud service component is the number of times that component sends data to other cloud service components per unit of time, or the number of times it calls other cloud service components. Modularity: Also known as modularity metric, it's a commonly used method to measure the strength of a network community structure. In other words, modularity measures the tightness of a community. Vertical scaling: This refers to enhancing the performance of a single machine's hardware, such as increasing the number of CPU cores (e.g., 32 cores), upgrading to a better network card (e.g., 10 Gigabit), upgrading to a better hard drive (e.g., SSD), expanding hard drive capacity (e.g., 2TB), and expanding system memory (e.g., 128GB). Horizontal scaling: This refers to increasing storage and computing power by adding more servers or program instances to distribute the load; for example, increasing the number of storage devices. Horizontal scaling involves adding more equivalent functional components in parallel to distribute the load. Vertical scaling, on the other hand, is the process of increasing the capacity of a given instance, i.e., expanding the capability of a point to handle more requests. As major vendors increasingly focus on the processing capacity of cloud platforms or cloud services under heavy load, a common method for stress-handling cloud service components within a cloud platform includes: operations personnel monitoring the out-degree and in-degree of each cloud service component and the data transmission frequency of the links between these components, and using experience to determine whether the cloud platform has hotspots or overloaded areas.Hot links. When hot spots and hot links exist on the cloud platform, operations and maintenance personnel scale up the capacity of these hot spots and links based on experience. For example, in a relationship graph consisting of multiple cloud service components' call relationships, there are cloud service component A and cloud service component B; cloud service component B is used to store data sent by cloud service component A. If operations and maintenance personnel determine that cloud service component A is a hot spot, they scale up cloud service component A to enable it to handle business data under the current load pressure. However, when the cloud platform is under heavy load, scaling up only cloud service component A, which is the hot spot, although improving the processing efficiency of cloud service component A, increases the load pressure on cloud service component B, making cloud service component B a new hot spot. Therefore, the above method cannot quickly adjust the load pressure of the cloud platform under heavy load, resulting in limited cloud platform performance. Based on this, this application provides a cloud platform management method. This method divides multiple cloud service components into multiple communities, including a first community, according to the calling relationships between them. The calling relationships between cloud service components within a community are relatively close, while the calling relationships between communities are relatively close. When the first community meets preset conditions, a second community identical to the first community is created (e.g., cloning the first community to obtain the second community), so that the second community shares some of the load with the first community. Therefore, when the cloud platform is under heavy load, by cloning the entire community composed of multiple cloud service components, the high degree of calling relationships between the cloud service components within the entire community allows for load balancing between the first and second communities, which can quickly adjust the load pressure of the cloud platform, thereby improving the performance of the cloud platform. This application provides a cloud platform management method applicable to the cloud platform system shown in Figure 1. The cloud platform system includes a cloud platform and a management device. The cloud platform provides powerful computing services to users. It comprises N cloud services with dependencies between them. One of the N cloud services provides a required service (e.g., storage service or gateway service) to the cloud platform. Each cloud service includes X cloud service components with dependencies between them, and one of the X cloud service components provides a sub-service to the cloud service. It should be noted that the number of cloud service components included in any two of the N cloud services can be different. The management device executes the cloud platform management method provided in this application to manage the cloud platform, thereby enabling the cloud platform to function properly.When the platform is under heavy load, the management device quickly reduces the load pressure on the cloud in the cloud platform. The management device executes the cloud platform management method provided in this application embodiment as shown in S110-S140 below, which will not be repeated here. It should be understood that the management device can be a cloud server in the cloud platform or a cloud server independent of the cloud platform. The specific location of the management device is not limited in this application embodiment. It should be noted that the system architecture and application scenarios described in this application embodiment are for the purpose of more clearly illustrating the technical solutions of this application embodiment and do not constitute a limitation on the technical solutions provided in this application embodiment. Those skilled in the art will know that with the evolution of system architecture and the emergence of new business scenarios, the technical solutions provided in this application embodiment are also applicable to similar technical problems. This application embodiment provides a cloud platform management method, which is applied to the management device in the cloud platform system shown in Figure 1; as shown in Figure 2, the method includes: S110-S140. S110: The management device obtains the calling relationship between multiple cloud service components in the cloud platform. The aforementioned cloud platform, as shown in Figure 1, comprises multiple cloud service components that interact with each other. These interaction relationships include one or more of the following: call information of the multiple cloud service components, the call frequency of the link between any two cloud service components, and the out-degree and in-degree of each cloud service component. The call information includes: when the first component is the primary caller, the identifiers of other cloud service components called by the first component; and when the first component is the called component, the identifiers of the cloud service components calling the first component; the first component can be any one of the multiple cloud service components. In other words, the call information includes the correspondence between the primary caller and the called component among the multiple cloud service components. For example, when cloud service component A calls cloud service component B, the primary caller is cloud service component A, and the called component is cloud service component B, i.e., cloud service component A → cloud service component B. The call frequency of the link is used to indicate the number of times data is transmitted per unit time. For example, the above-mentioned call relationship can be represented by the call relationship graph shown in Figure 3. Each node in this graph (e.g., ②) represents a cloud service component (e.g., cloud service component 2). Any node in this graph (e.g., node 2 representing cloud service component 2) includes: the identifiers of other cloud service components called when cloud service component 2 acts as the caller, such as ②-①; and the identifiers of the cloud service components that call cloud service component 2 when cloud service component 2 acts as the called party, such as ⑧-② and ⑥-②. It also includes: the call frequency of the links between cloud service component 2 and the cloud service components that have a call relationship with it, such as the call frequency of link 1 between ② and ⑥; and also includes: the out-degree and in-degree of cloud service component 2.It should be noted that the specific implementation process of the above S110 may be that the management device obtains the above call relationship from the cloud platform through a monitoring service (such as a network management service), or the user may pre-import the above call relationship into the management device, and then the management device directly obtains the call relationship from the local machine; the specific implementation method of the above S110 is not limited in this application embodiment. S120, the management device divides multiple cloud service components into multiple communities based on the calling relationships between them. The degree of closeness of the calling relationship between cloud service components within any one of these communities is greater than or equal to a preset closeness. Conversely, the degree of closeness between a cloud service component in one of these communities and cloud service components in other communities is less than the preset closeness; that is, the degree of closeness of the calling relationship between any two communities is less than the preset closeness. The closeness of the calling relationship is used to characterize the calling frequency between two cloud service components. In other words, cloud service components with a high degree of call relationship among multiple cloud service components are divided into one community, while cloud service components with a low degree of call relationship among multiple cloud service components are divided into different communities. The specific implementation of the above S120 can be that the management device divides the above multiple cloud service components into multiple communities based on the community discovery algorithm and the call relationship between the above multiple cloud service components; or the management device can divide the above multiple cloud service components into multiple communities based on a preset training model. The specific implementation of the above S120 is not limited in this application embodiment. It should be noted that when the above S120 is a management device that divides the above multiple cloud service components into multiple communities based on the community discovery algorithm, its specific implementation method includes: treating each cloud service component as an independent community; then, based on the following formula (1), calculating the modularity Q of each community in the multiple communities; when it is necessary to determine whether to divide community A and community B into one community, calculating the modularity Q of community C obtained after dividing community A and community B into one community; when the difference between the modularity Q of community C and the modularity Q of community A is greater than 0, dividing community A and community B into one community; when the difference between the modularity Q of community C and the modularity Q of community A is less than or equal to 0, not dividing community A and community B into one community. Repeating this process until the modularity Q of any two communities (e.g., community 1 and community 2) obtained by merging the communities (e.g., community 3) is greater than the modularity Q of community 1 and the modularity Q of community 2.Where m is the sum of the weights of all edges in the call network corresponding to the above call relationship; £in is the sum of the weights of all edges between nodes within community A; stot is the sum of the weights of all edges connected to nodes in community A, i.e., the weights of all edges within community A (abbreviated as: weights of internal edges in community A) and the weights of all edges between nodes in other communities and nodes in community A (abbreviated as: weights of external edges in community A). It should be noted that the number of edges between one node and another node is used to represent the number of calls between that node and that other node per unit time. For example, assuming the call network corresponding to the above call relationship is shown in Figure 3, based on the community detection algorithm, the call network is divided into three communities, as shown in Figure 4. Nodes 6-10 are divided into the first community, nodes 1-5 into the second community, and nodes 11-15 into the third community. Among these three communities, the edge connections between multiple nodes within any one community are relatively close, while the edge connections between any two communities are relatively close. S130. The management device determines whether the target component within the first community meets the first preset condition. It should be understood that the first community is any one of the multiple communities, and the first community includes multiple cloud service components. It should be noted that the higher the load of the first community, the lower the performance of the target component within the first community; based on this, the first preset condition is used to indicate that the performance of the target component is lower than a preset performance. In one implementation, the target component may be a cloud service component among the multiple cloud service components in the first community that meets a second preset condition; wherein the second preset condition includes: the top X cloud service components with the highest popularity value among the multiple cloud service components in the first community, or the top μ cloud service components with the lowest popularity value among the multiple cloud service components in the first community; wherein the popularity value of a cloud service component is the sum of the out-degree and in-degree of the cloud service component, and X and m are both integers greater than 0. In other words, the management device calculates the popularity value of all cloud service components within the first community and sorts them by popularity value. Then, it determines the top X cloud service components with the highest popularity value as target components, or the top m cloud service components with the lowest popularity value as target components. For example, suppose the first community includes 5 cloud service components, namely cloud service component A to cloud service component ε; the popularity values ​​of these 5 cloud service components, sorted from largest to smallest, are: cloud service component B, cloud service component C, cloud service component E, cloud service component D, and cloud service component A; and suppose the second preset condition includes: the top 2 cloud service components with the highest popularity value among the multiple cloud service components in the first community; then, the target components are cloud service component B and cloud service component C. In another implementation, the target components are at least one cloud service component specified by the user among the multiple cloud service components in the first community.Service component; that is, the above-mentioned target component is pre-specified by the user in the first community. 7 5 10 15 20 25 30 35 40 45 WO 2024 / 227365 PCT / CN2024 / 071300 It should be noted that the above-mentioned first preset condition is used to determine whether the overall performance of the above-mentioned first community has reached the upper limit. The specific implementation of the above-mentioned S130 includes: when the management device determines that the first target feature of the above-mentioned target component meets the above-mentioned first preset condition; wherein, the first target feature is at least one feature specified by the user among multiple features of the above-mentioned target component. In one embodiment, the above-mentioned first target feature includes: at least one of the following: processor (central processing unit, CPU) utilization, memory utilization, and input / output (I / O) throughput. When the first target feature includes CPU utilization, the first preset condition includes: CPU utilization is greater than a preset CPU utilization (i.e.: preset first CPU utilization). For example: the first preset condition includes CPU utilization greater than 90%. When the first target feature includes memory utilization, the first preset condition includes: memory utilization greater than a preset memory utilization (i.e., a preset first memory utilization); for example, the first preset condition includes memory utilization greater than 80%. When the first target feature includes I / O throughput, the first preset condition includes: I / O throughput greater than a preset I / O throughput (i.e., a preset first I / O throughput); for example, the first preset condition includes memory utilization greater than 500 MB / s. For example, assuming the target component is cloud service component A; and the first target feature includes CPU utilization, memory utilization, and I / O throughput; then, the first preset condition includes: CPU utilization greater than a preset first CPU utilization, memory utilization greater than a preset first memory utilization, and I / O throughput greater than a preset first I / O throughput. At this point, assuming a preset first CPU utilization rate of 90%, a preset first memory utilization rate of 80%, and a preset first I / O throughput of 500 MB / s; and cloud service component A currently has a CPU utilization rate of 80%, a memory utilization rate of 91%, and an I / O throughput of 600 MB / s; then, since the current CPU utilization rate of cloud service component A is 80%, which is less than the preset first CPU utilization rate of 90%, the target component within the first community does not meet the first preset condition. When the first community does not meet the first preset condition, it indicates that the current load pressure of the first community is relatively low, so that the overall performance of the first community is higher than the lower limit of the community's performance, and therefore the management device ends the current method. When the target component within the first community meets the first preset condition, the management device executes the following S140.S140. The management device creates a second community. The cloud service components and their inter-component calling relationships included in the second community are the same as those in the first community. In other words, the second community is identical to the first community. S140 is implemented by cloning the first community to obtain the second community. Specifically, the management device obtains the orchestration template of the first community, which includes information such as component resource types, relationships between component resources, and the aforementioned calling relationships. The component resource types include the resource types required by each of the multiple cloud service components in the first community (e.g., storage or computing resources). The relationships between component resources indicate the relationships between the resources required by each of the multiple cloud service components in the first community (e.g., dependencies). Then, the management device deploys a cloned community identical to the first community, i.e., the second community, based on this orchestration template. It should be noted that the aforementioned orchestration template can be obtained by the management device after executing S120, acquiring information such as the component resource types, relationships between component resources, and the aforementioned call relationships of the first community, and storing it locally as an orchestration template; alternatively, the management device can directly obtain the orchestration template from the first community when executing S140. The format of the orchestration template is generally JSON or Extensible Markup Language (XML). This application embodiment does not specifically limit the method of obtaining the orchestration template. It should also be noted that the aforementioned second community is used to share some of the load of the first community; specifically, the second community can be used to share some of the load of some cloud service components in the first community, or it can be used to share some of the load of all cloud service components in the first community; the specific method is determined by the call relationships and load balancing strategies within the first community. It should be understood that the aforementioned second community is completely identical to the aforementioned first community. Based on a load balancing strategy, the second community is used to share part of the load of the first community. For example, assuming the load balancing strategy is a round-robin strategy, the second community will share half of the load of the first community. For example, suppose the first community includes: cloud service component A - cloud service component C; where cloud service component A calls cloud service component B; cloud service component B calls cloud service component C; that is: cloud service component A - cloud service component B - cloud service component C. In this case, if cloud service component D outside the first community (such as the third community) calls cloud service component β, since cloud service component β calls cloud service component C but not cloud service component A, in this scenario, when cloud service component D sends multiple...When sending multiple sets of data, cloud service components β and C in the second community share some of the load of cloud service components B and C based on the load balancing strategy. In another example, based on the above example, the first community includes: cloud service component A, cloud service component B, and cloud service component C. 8 WO 2024 / 227365 PCT / CN2024 / 071300 5 10 15 20 25 30 35 40 45 In this case, if cloud service component D outside the first community (such as the third community) calls cloud service component A, since cloud service component A calls cloud service component B, and cloud service component B calls cloud service component C; therefore, in this scenario, when cloud service component D sends multiple sets of data to cloud service component A, cloud service component A, and cloud service component C in the second community share some of the load of cloud service component A, cloud service component B, and cloud service component C in the first community based on the load balancing strategy. In one embodiment, when the overall performance of the first community and the second community is lower than the user's preset conditions, the management device releases the first community or the second community, thereby saving available resources in the cloud platform. The specific method for releasing the first community or the second community is described in the prior art and will not be repeated here. Based on this, this application provides a cloud platform management method. This method divides multiple cloud service components into multiple communities, including the first community, according to the calling relationships between these components. The calling relationships between cloud service components within a community are relatively close, while the calling relationships between communities are relatively close. When the first community meets preset conditions, a second community identical to the first community is created (e.g., the first community is cloned to obtain the second community), so that the second community shares some of the load of the first community. Therefore, when the cloud platform is under heavy load, by cloning the entire community composed of multiple cloud service components, the high degree of calling relationships between the cloud service components within the entire community allows for load balancing between the first community and the second community, which can quickly adjust the load pressure of the cloud platform and improve its performance. It should be noted that there are calling relationships between communities. When the calling relationship between one community (such as the first community) and another community is very close, scaling up only the first community may lead to excessive load on the other community, resulting in poor load reduction for the entire cloud platform. Therefore, in one embodiment, referring to Figure 2 and Figure 5, the specific implementation of S140 above further includes: S141-S144. S141: The management device obtains the modularity of the first community and each of the other communities among the multiple communities excluding the first community. Modularity describes the closeness of the calling relationship between cloud service components in one community and cloud service components in another community.The module degree is a value used to describe the tightness of the calling relationship between communities. It should be noted that the specific implementation of S141 above can be that after the management device executes S120 above, it saves the module degree of any one community among multiple communities and other communities, and directly obtains the module degree of the first community and other communities from the local machine when the management device executes S141 above; the specific implementation of S141 above can also be that the management device calculates the module degree of the first community and other communities according to the formula (1) above; the specific implementation method of S141 above is not limited in this application embodiment. It should be noted that when the specific implementation method of S141 above is that the management device calculates the module degree of the first community and other communities according to the formula (1) above, the specific calculation process refers to the relevant description of S120 above, and will not be repeated here. S142. The management device determines whether there is a target community among other communities based on the module degree of the first community and other communities. The target community is the community among the other communities whose modularity with the first community is greater than a threshold. That is, the closeness of the call relationship between the target community and the first community is greater than the threshold; in other words, the target community is the community among the other communities with a relatively close call relationship with the first community. For example, suppose the above multiple communities include four communities, namely the first to the fourth community; wherein the modularity between the first and second communities is 0.6; the modularity between the first and third communities is 0.2; the modularity between the first and fourth communities is 0.3; and the threshold is 0.55. In this case, the modularity of the first and second communities (0.6) is greater than the threshold (0.55), so the second community is determined as the target community. When there is no target community among the other communities, execute S143 below. When there is a target community among the other communities, execute S144 below. S143: The management device creates the second community. It should be noted that the implementation of S143 is consistent with that of S140. For a detailed description of S143, please refer to the relevant description of S140 above; it will not be repeated here. S144. The management device creates a second community combination. The specific implementation of S144 includes: obtaining a second community combination by cloning the first community combination, which includes the first community and the target community. The second community combination is the same as the first community combination, which includes the first community and the target community; that is, the second community combination is a clone of the first community combination. The second community combination includes a second community and a third community, where the second community is a clone of the first community; the third community is a clone of the target community. In other words, the specific implementation of S144 is to clone the first community and the target community as a whole to obtain the second community combination. 9 WO 2024 / 227365 PCT / CN2024 / 071300 5 10 15 20 25 30 35It should be understood that the invocation relationship between the second community and the third community in the foregoing second community combination is consistent with the invocation relationship between the first community and the target community in the foregoing first community combination. It should be noted that the foregoing second community combination is configured to share part of the load of the first community combination, wherein the second community combination may specifically be configured to share part of the load of some cloud service components in the first community combination, and may also be configured to share part of the load of all cloud service components in the first community combination; a specific sharing manner is similar to the sharing manner in the foregoing S140, and for details, reference may be made to relevant descriptions in S140, which are not repeated herein. It should be noted that an implementation manner of the foregoing S144 is similar to an implementation manner of S140, and for a specific description of S144, reference may be made to the foregoing related description of S140, which is not repeated herein. According to the embodiments of this application, by determining whether there is a target community with a modularity greater than a threshold with a first community among multiple communities, when the target community exists (that is, there is a community with a relatively high closeness of invocation relationship with the first community), the first community and the target community are taken as an overall community (that is, a first community combination), and a second community combination that is the same as the first community combination and includes a second community and a third community is created; the second community combination is used to share the load of the first community combination, thereby solving the problem of excessive load pressure on the target community caused by only expanding the capacity of the first community, and further improving the intensity of reducing the load pressure of a cloud platform. It should be noted that the foregoing S130-S140 implement capacity expansion at the community level (that is, capacity expansion by means of community cloning) when a target component in the first community satisfies a first preset condition; and when the target component in the first community does not satisfy the first preset condition, there may also be a problem that at least one cloud service component and a link have a relatively large load in the first community, which may cause the entire cloud platform to crash in severe cases. Based on this, in an embodiment, when the management device determines in the foregoing S130 that the target component in the first community does not satisfy the first preset condition, as shown in FIG. 6, the foregoing method further includes: S210-S220. S210. The management device determines hot nodes and hot links according to invocation relationships between cloud service components in the first community. The foregoing hot nodes are cloud service components with a relatively high invocation frequency (that is, heat value) among multiple cloud service components in the first community, wherein the heat value of a cloud service component is a sum of an out-degree of the cloud service component and an in-degree of the cloud service component. The foregoing hot links are links with a relatively high invocation frequency among multiple cloud service components in the foregoing first community. In an example, the manner for determining the hot nodes and the hot links is as follows: A formula for determining the hot nodes in the foregoing first community is shown in the following formula (2). Node = max{Σ rij + Σ rji} (2) where, Node represents a hot node; max represents taking a maximum value; rij represents the number of times that a node i (that is, cloud service component i) invokes a node j; xX is the total number of nodes (i.e., cloud service components) in the first community. The above hot link is the link with a high call frequency between multiple cloud service components in the first community; the formula for determining the hot link is shown in the following formula (3). Link = max{£*i,j=i rij} (3) Wherein, Link is the hot link; max means taking the maximum value; rij means the number of times node i calls node j, that is: the call frequency of the link used when node i calls node j; X is the total number of nodes in the first community. For example, in the matrix data corresponding to the first community shown in Figure 7, rij(x) means that the number of times node i calls node j is X times; The first community includes 5 nodes, where each node represents a cloud service component. The specific data called by these 5 nodes includes {rll(2), rl2(l), rl3(2), rl4(2), rl5(2)}, {r21(l), r22(l), r23(5), r24(3), r25(8)}, {r31(3) > r32(4), r33(0), r34(2), r35(6)}, {r41(2), r42(l), r43(3), r44(0), r45(2)}, and {r51(7), r52(2), r53(2), r54(9), r55(5)}; Based on the above formula (2), the heat values ​​of node 1 to node 5 are calculated respectively, and the heat values ​​of node 1 to node 5 are as follows: 22, 26, 27, 24, 43; that is, the node with the highest heat value 5 is identified as a hot spot, and the link corresponding to r54 with the highest call frequency is identified as a hot link. In another example, the specific implementation method of the above S210 includes: S1-S3. S1, the management device obtains the heat value of each of the multiple cloud service components in the first community and the call frequency of multiple links between the multiple cloud service components. It should be noted that the above management device calculates the heat value of each of the multiple cloud service components based on the above formula (2); the above management device determines the call frequency of the above multiple links from the call relationship of the cloud service components in the first community. S2, the management device identifies the top N cloud service components with the highest heat value as hot spots. 10 5 10 15 20 25 30 35 WO 2024 / 227365 PCT / CN2024 / 071300 The above N is a preset integer greater than 0. For example, based on the example of S210 above; assuming N is 2, since the heat values ​​of nodes 1-5 are 22, 26, 27, 24, 43 respectively; then, the management device will identify nodes 5 and 3 as hotspots. S3, the management device will identify the top m links with the highest call frequency as hot links. M is a preset integer greater than 0.For example, based on the example of S210 above; assuming m is 2, then the management device will identify the link corresponding to r54 (hereinafter referred to as r54) and rl5 as hot links. It should be noted that after the management device identifies the hotspots and hot links in the first community, it can mark the hotspots and hot links in the call network corresponding to the first community, so that when users view the call network, they can intuitively understand the distribution of hotspots and hot links in the first community, improving the user experience. S220, the management device processes the hotspots and hot links according to the preset reduction strategy. The preset reduction strategy is used to make the processed hotspots and hot links meet their respective current load requirements; that is, the preset reduction strategy is a method to reduce the load pressure of hotspots and hot links. The specific implementation of S220 includes two methods, as follows: In one implementation method, the specific implementation method of S220, as shown in Figure 8, includes: S220a-S220d. S220a, the management device determines the target hotspot reduction strategy from the first target correspondence based on the identifier of the community where the hotspot is located and the second target characteristics of the hotspot. The aforementioned second target feature is a feature pre-specified by the user from among multiple features of the aforementioned hotspot. For example, the second target feature includes at least one of CPU utilization, memory utilization, and I / O throughput. It should be understood that the aforementioned second target feature is a feature pre-specified by the user from among multiple features of the hotspot based on the specific function of the hotspot (i.e., the type of service provided to the cloud platform). For example, when the hotspot is a cloud service component used for data storage, the second target feature may include storage space utilization; or when the hotspot is a cloud service component used for data computation, the second target feature may include CPU utilization. It should be noted that this application embodiment uses the second target feature including CPU utilization, memory utilization, and I / O throughput as an example for illustration, and will not be elaborated further thereafter. The aforementioned first target correspondence includes the identification of multiple communities, the correspondence between multiple third preset conditions and multiple hotspot reduction strategies; wherein, the first target correspondence is shown in Table 1 below, and the identification of the multiple communities includes: the first community and the second community. The various third preset conditions include: "CPU utilization > preset second CPU utilization, memory utilization > preset second memory utilization, I / O throughput > preset second I / O throughput"; "CPU utilization > preset third CPU utilization, memory utilization > preset third memory utilization, I / O throughput > preset third I / O throughput"; "CPU utilization > preset fourth CPU utilization, I / O throughput > preset fourth I / O throughput". These multiple hotspot reduction strategies include: vertical scaling and horizontal scaling. The aforementioned third preset conditions are conditions that the above-mentioned second target characteristics must meet, and these third preset conditions are used to determine whether to expand the capacity of the aforementioned hotspots. Table 1 Community IdentifierThe third preset condition hotspot reduction strategy: First community CPU utilization > preset second CPU utilization, memory utilization > preset second memory utilization, I / O throughput (preset second I / O throughput). Horizontal expansion: First community CPU utilization > preset third CPU utilization, memory utilization > preset third memory utilization, I / O throughput ν (preset third I / O throughput). Vertical expansion: Second community CPU utilization > preset fourth CPU utilization, I / O throughput ν (preset fourth I / O throughput). The above vertical expansion provides concurrency by improving the performance of single-machine hardware, specifically by increasing the number of CPUs in the hotspot, and / or increasing the memory capacity in the hotspot (e.g., adding memory modules), and / or increasing the data bus of the hotspot. The above horizontal expansion increases the number of cloud service components represented by the hotspot and reduces the load pressure of the hotspot through load balancing strategies; for example, when the hotspot is cloud service component A, increase the number of cloud services for component A. It should be noted that the specific implementation methods of the above vertical and horizontal expansion refer to existing technologies and will not be repeated here. 30 35 40 WO 2024 / 227365 PCT / CN2024 / 071300 The identifier of the community corresponding to the above-mentioned target hotspot reduction strategy is consistent with the identifier of the first community where the hotspot is located; and the second target feature of the hotspot satisfies the third preset condition corresponding to the above-mentioned target hotspot reduction strategy. For example, assuming that the identifier of the community where the hotspot is located is the first community, the CPU utilization rate of the hotspot is greater than the preset third CPU utilization rate, memory utilization rate is greater than the preset third memory utilization rate, and I / O throughput is greater than the preset third I / O throughput; then, the target hotspot reduction strategy of the hotspot is vertical expansion. S220b. The management device processes the hotspot according to the target hotspot reduction strategy. For example, based on the example in S220a above, the hotspot is vertically expanded, specifically including: increasing the number of CPUs in the hotspot, increasing the memory storage of the hotspot, and increasing the number of data buses in the hotspot. S220c. The management device determines the target hotspot reduction strategy from the second target correspondence based on the third target feature of the hotspot. The aforementioned third target feature is a feature pre-specified by the user among multiple features of the aforementioned hot link; for example, the third target feature includes: data transmission success rate and I / O throughput. It should be noted that this embodiment uses I / O throughput as an example of the third target feature for illustration, and will not be repeated hereafter. The aforementioned fourth preset condition is a condition that the aforementioned third target feature must satisfy, and this fourth preset condition is used to determine whether to extend the aforementioned hot link. The aforementioned second target correspondence includes the correspondence between multiple fourth preset conditions and multiple hot link mitigation strategies; the specific correspondence is shown in the table below.As shown in Table 2, the multiple fourth preset conditions include: "I / O throughput < 600M / s, and I / O throughput 2400M / s" and "I / O throughput V 400M / s"; the multiple hot link reduction strategies include horizontal expansion and vertical expansion. Table 2 Fourth Preset Condition Hot Link Reduction Strategies I / O throughput < 600M / s, and I / O throughput % 00M / s Horizontal Expansion I / O throughput < 400M / s Vertical Expansion The above horizontal expansion can specifically increase the number of links so that part of the load of the hot link is distributed to the added links. The above vertical expansion can specifically increase the bandwidth of the hot link so that the amount of data that can be transmitted is increased. The specific implementation of S220c includes: the management device determines the hot link reduction strategy corresponding to the fourth preset condition that the third target characteristic of the hot link is satisfied as the target hot link reduction strategy. For example, assuming the current hot link's I / O throughput is 550M / s, then the hot link reduction strategy (i.e., horizontal scaling) corresponding to the fourth preset condition of I / O throughput < 600M / s and I / O throughput < 00M / s will be determined as the target hot link reduction strategy. S220d: The management device processes the hot link according to the target hot link reduction strategy. For example, based on the example in S220c above, the target hot link reduction strategy is horizontal scaling, thereby increasing the number of hot links so that the added links share part of the load of the hot link. In the above embodiment, when the target component in the first community does not meet the first preset condition, hotspots and hot links in the first community are determined. Then, based on the community where the hotspot is located and the third preset condition satisfied by the second target feature of the hotspot, a target hotspot reduction strategy is determined, and the hotspot is processed based on the target hotspot reduction strategy to reduce the load pressure of the hotspot. Similarly, the management device determines the hotspot reduction strategy corresponding to the fourth preset condition satisfied by the third target feature of the hot link as the target hotspot reduction strategy, and processes the hotspot according to the target hotspot reduction strategy to reduce the load pressure of the processed hotspot, thereby solving the problem of cloud platform downtime caused by the presence of at least one cloud service component and link with a large load in the first community. In another implementation, the specific implementation method of the above S220, as shown in Figure 9, includes: S220A-S220C. S220A: The management device determines the target type template from multiple preset type templates based on the distribution characteristics of the hotspots and hot links. The distribution characteristics of the hotspots and hot links are used to characterize the location information of the hotspots and hot links in the space of the first community; that is, The aforementioned distribution characteristics are used to indicate the location information of the aforementioned hotspots and hot links within the first community. The node and link characteristics differ between any two of the aforementioned preset type templates, and the aforementioned target type template...The type template with the highest matching degree among the above-mentioned multiple preset type templates is the one that matches the distribution characteristics of hotspots and hot links. The specific implementation method of the above-mentioned S220A includes: the management device calculates the matching degree with each preset type template according to the distribution characteristics of hotspots and hot links; then, the management device determines the template type with the highest matching degree as the target type template. It should be noted that the above-mentioned management device can calculate the matching degree between the distribution characteristics of hotspots and hot links and each preset type template based on a training model; the above-mentioned management device can also calculate the matching degree between the distribution characteristics of hotspots and hot links and each preset type template based on other existing algorithms. Specifically, the specific implementation method of the above-mentioned S220A is not limited in this application embodiment. S220B, the management device determines the target reduction strategy corresponding to the target type template from the correspondence between multiple type templates and multiple reduction strategies. The above-mentioned multiple template types include various different types of templates, such as centralized templates and distributed templates. The above correspondence is shown in Table 3 below. The above-mentioned multiple template types include: Type Template 1, Type Template 2, and Type Template 3; the above-mentioned multiple mitigation strategies include: expansion hotspot, expansion hotspot and expansion hot link, and expansion hot link. Table 3 Type Template Mitigation Strategies Type Template 1 Expansion Hotspot Type Template 2 Expansion Hotspot and Expansion Hot Link Type Template 3 Expansion Hot Link It should be noted that the specific method of the above-mentioned expansion hotspot can be vertical expansion or horizontal expansion; the specific method of the above-mentioned expansion hot link can be vertical expansion or horizontal expansion. In this application embodiment, the specific methods of the above-mentioned expansion hotspot and expansion hot link are not limited. For example, Type Template 1 in Table 1 is the centralized template in Figure 10, Type Template 2 is the link template in Figure 10, and Type Template 3 is the distributed template in Figure 10; when the above-mentioned target type template is a centralized template, the expansion hotspot is determined as the target mitigation strategy. S220C. The management device processes hotspots and hot links according to the target mitigation strategy. In the above embodiments, when the target component within the first community does not meet the first preset condition, hotspots and hot links within the first community are identified. Then, the management device determines a target type template from multiple preset type templates based on the distribution characteristics of the hotspots and hot links, and processes the hotspots and hot links according to the target reduction strategy corresponding to the target type template, thereby reducing the load pressure on the processed hotspots and hot links. This solves the problem of cloud platform downtime caused by the presence of at least one cloud service component and link with a high load in the first community. The above mainly describes the solution provided by the embodiments of this application from a methodological perspective. To achieve the above functions, the management device includes...The hardware structure and / or software modules corresponding to each function are described. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application. Embodiments of this application can, exemplarily, divide the management device into functional modules according to the above method. For example, the management device may include functional modules corresponding to each functional division, or two or more functions may be integrated into a single processing module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in the embodiments of this application is illustrative and only represents a logical functional division; other division methods may exist in actual implementation. In the case of dividing each functional module according to each function, Figure 11 shows a possible structural schematic diagram of the management device involved in the above embodiments. As shown in Figure 11, the management device includes a partitioning module 1101 and a creation module 1102. The partitioning module 1101 is used to partition multiple cloud service components into multiple communities based on the calling relationships between multiple cloud service components in the cloud platform; for example, executing step S1200 in the above method embodiment. The creation module 1102 is used to create a second community if the target component in the first community among the multiple communities meets a first preset condition; for example, executing step S140 in the above method embodiment. Optionally, the partitioning module 1101 is used to partition multiple cloud service components into multiple communities based on a community discovery algorithm and calling relationships. Optionally, the creation module 1102 is used to create a second community if the first target feature of the target component meets the first preset condition. Optionally, the management device further includes an acquisition module 1103 and a determination module 1104. The acquisition module 1103 is used to acquire the modularity of the first community and each of the other communities in the plurality of communities excluding the first community; for example, executing step S14L in the above method embodiment. The determination module 1104 is used to determine whether the target community exists in the other communities based on the modularity; for example, executing step S142 in the above method embodiment. The creation module 1102 is used to create a second community when the target community does not exist in the other communities; for example, executing step S143 in the above method embodiment. Optionally, the creation module 1102 is also used to create a second community combination when the target community exists in the other communities; for example, executing step S143 in the above method embodiment.Step S144 in the method embodiment. Optionally, the determining module 1104 is used to determine hotspots and hot links based on the call relationship between cloud service components in the first community if the target component does not meet the first preset condition; for example, executing step S210 in the above method embodiment; the determining module 1104 is used to process hotspots and hot links according to a preset reduction strategy; for example, executing step S220 in the above method embodiment. Optionally, the determining module 1104 is used to determine a target hotspot reduction strategy from the first target correspondence based on the identifier of the community where the hotspot is located and the second target feature of the hotspot; for example, executing step S220a0 in the above method embodiment. The determining module 1104 is used to process the hotspot according to the target hotspot reduction strategy; for example, executing step S220b in the above method embodiment. Optionally, the determining module 1104 is used to determine a target hot link reduction strategy from the second target correspondence based on the third target feature of the hot link; for example, executing step S220c in the above method embodiment. The determining module 1104 is further configured to process hot links according to the target hot link reduction strategy; for example, executing step S220d in the above method embodiment. Optionally, the determining module 1104 is configured to determine a target type template from multiple preset type templates based on the distribution characteristics of hot spots and hot links; for example, executing step S220A in the above method embodiment. The determining module 1104 is configured to determine the target reduction strategy corresponding to the target type template from the correspondence between multiple type templates and multiple reduction strategies; for example, executing step S220B in the above method embodiment. The determining module 1104 is further configured to process hot spots and hot links according to the target reduction strategy; for example, executing step S220C in the above method embodiment. The partitioning module 1101, creation module 1102, acquisition module 1103, and determination module 1104 can all be implemented in software or hardware. For example, the implementation of partitioning module 1101 will be described below. Similarly, the implementation of creation module 1102, acquisition module 1103, and determination module 1104 can refer to the implementation of partitioning module 1101. As an example of a software functional unit, partitioning module 1101 may include code running on a computing instance. A computing instance may include at least one of a physical host (computing device), a virtual machine, or a container. Further, the aforementioned computing instance may be one or more. For example, partitioning module 1101 may include code running on multiple hosts / virtual machines / containers. It should be noted that the multiple hosts / virtual machines / containers used to run the code can be distributed in the same region or in different regions. Further, the multiple hosts / virtual machines / containers used to run the code can be distributed in the same availability zone (AZ).In this context, virtual machines (VMs) can be distributed across different Availability Zones (AZs), each AZ comprising one or more geographically proximate data centers. Typically, a region can include multiple AZs. Similarly, multiple hosts / VMs / containers used to run the code can be distributed within the same Virtual Private Cloud (VPC) or across multiple VPCs. Typically, a VPC is set up within a region. Communication between two VPCs within the same region, and between VPCs in different regions, requires a communication gateway within each VPC to interconnect them. As an example of a hardware functional unit, partitioning module 1101 can include at least one computing device, such as a server. Alternatively, partitioning module 1101 can also be a device implemented using an application-specific integrated circuit (ASIC) or a programmable logic device (PLD). The aforementioned PLD can be implemented using a complex programmable logical device (CPLD), a field-programmable gate array (FPGA), generic array logic (GAL), or any combination thereof. The multiple computing devices included in partitioning module 1101 can be distributed in the same region or in different regions. Similarly, the multiple computing devices included in partitioning module 1101 can be distributed in the same Availability Zone (AZ) or in different AZs. Likewise, the multiple computing devices included in partitioning module 1101 can be distributed in the same Virtual Private Cloud (VPC) or in multiple VPCs. These multiple computing devices can be any combination of computing devices such as servers, ASICs, PLDs, CPLDs, FPGAs, and GALs. It should be noted that, in other embodiments, the partitioning module 1101 can be used to execute any step in the above-described cloud platform management method, the creation module 1102 can be used to execute any step in the above-described cloud platform management method, the acquisition module 1103 can be used to execute any step in the above-described cloud platform management method, and the determination module 1104 can be used to execute any step in the above-described cloud platform management method. The steps implemented by the partitioning module 1101, creation module 1102, acquisition module 1103, and determination module 1104 can be specified as needed. The above 14 WO are implemented by the partitioning module 1101, creation module 1102, acquisition module 1103, and determination module 1104 respectively.2024 / 227365 PCT / CN2024 / 071300 5 10 15 20 25 30 35 40 The different steps in the cloud platform management method described above are used to implement all the functions of the management device. This application also provides a computing device cluster. The computing device cluster includes at least one computing device. The computing device can be a server, such as a central server, an edge server, or a local server in a local data center. In some embodiments, the computing device can also be a desktop computer, a laptop computer, or a smartphone, etc. As shown in FIG12, the computing device cluster includes at least one computing device 100. The memory 106 of one or more computing devices 100 in the computing device cluster can store the same instructions for executing the cloud platform management method described above. In some possible implementations, the memory 106 of one or more computing devices 100 in the computing device cluster can also separately store partial instructions for executing the cloud platform management method described above. In other words, a combination of one or more computing devices 100 can jointly execute the instructions for executing the cloud platform management method described above. It should be noted that the memories 106 in different computing devices 100 within the computing device cluster can store different instructions, each used to execute a portion of the management device's functions. That is, the instructions stored in the memories 106 of different computing devices 100 can implement the functions of one or more modules among the partitioning module 1101, creation module 1102, acquisition module 1103, and determination module 1104. In some possible implementations, one or more computing devices in the computing device cluster can be connected via a network. This network can be a wide area network (WAN) or a local area network (LAN), etc. Figure 13 illustrates one possible implementation. As shown in Figure 13, two computing devices 100 and 1008 are connected via a network. Specifically, they are connected to the network through the communication interfaces in each computing device. In this type of possible implementation, the memory 106 in computing device 100A stores instructions for executing the functions of the partitioning module 1101 and creation module 1102. Simultaneously, the memory 106 in computing device 100B stores instructions for executing the functions of the acquisition module 1103 and determination module 1104. The connection method between the computing device clusters shown in Figure 13 can be considered because the management method of the cloud platform provided in this application needs to obtain the modularity of the first community and each of the other communities besides the first community, and determine the target community from the other communities. Therefore, it is considered that the functions implemented by the acquisition module 1103 and the determination module 1104 are performed by the computing device 100B. It should be understood that the functions of the computing device 100A shown in Figure 13 can also be performed by multiple computing devices 100. Similarly, the functions of the computing device 100B can also be performed by multiple computing devices 100.This application embodiment also provides another computing device cluster. The connection relationship between the computing devices in this computing device cluster can be similarly referred to the connection method of the computing device cluster described in Figures 4 and 5. The difference is that the memory 106 of one or more computing devices 100 in this computing device cluster can store the same instructions for executing the management method of the cloud platform. In some possible implementations, the memory 106 of one or more computing devices 100 in this computing device cluster can also store partial instructions for executing the management method of the cloud platform. In other words, a combination of one or more computing devices 100 can jointly execute the instructions for executing the management method of the cloud platform. This application embodiment also provides a computer program product containing instructions. The computer program product can be a software or program product containing instructions that can run on a computing device or be stored in any usable medium. When the computer program product runs on at least one computing device, it causes at least one computing device to execute the management method of the cloud platform. This application embodiment also provides a computer-readable storage medium. The computer-readable storage medium can be any usable medium that a computing device can store or a data storage device such as a data center containing one or more usable media. The usable medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive). The computer-readable storage medium includes instructions that instruct a computing device to execute a cloud platform management method. Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the protection scope of the technical solutions of the embodiments of the present invention. 15 WO 2024 / 227365 PCT / CN2024 / 071300 5 10 15 20 25 30 35 40 45 Claim 1. A cloud platform management method, characterized in that it includes: dividing the multiple cloud service components into multiple communities according to the calling relationship between multiple cloud service components in the cloud platform; wherein each community includes at least two cloud service components from the multiple cloud service components, the degree of closeness of the calling relationship between the cloud service components included in each community is greater than or equal to a preset degree of closeness, and the degree of closeness between the cloud service components in any one of the multiple communities and the cloud service components in other communities is less than the preset degree of closeness;If a target component within a first community of the plurality of communities meets a first preset condition, a second community is created. The second community is identical to the first community and is used to share part of the load of the first community. The first preset condition indicates that the performance of the target component is lower than a preset performance. The first community is any one of the plurality of communities. The target component is at least one of the plurality of cloud service components in the first community. 2. The method according to claim 1, wherein the call relationship includes one or more of the following: call information of the plurality of cloud service components in the cloud platform, call frequency of the link between any two cloud service components in the plurality of cloud service components in the cloud platform, and out-degree and in-degree of any one cloud service component in the plurality of cloud service components in the cloud platform. 3. The method according to claim 1 or 2, wherein dividing the plurality of cloud service components into multiple communities based on the call relationship between the plurality of cloud service components in the cloud platform includes: dividing the plurality of cloud service components into the plurality of communities based on a community discovery algorithm and the call relationship. 4. The method according to any one of claims 1-3, wherein the target component comprises: a cloud service component among a plurality of cloud service components in the first community that satisfies a second preset condition, or at least one cloud service component specified by the user among a plurality of cloud service components in the first community. 5. The method according to claim 4, wherein the second preset condition comprises: the top X cloud service components with the highest popularity values ​​among a plurality of cloud service components in the first community; the popularity value is the sum of the out-degree and in-degree of the cloud service component, where X is an integer greater than 0. 6. The method according to any one of claims 1-5, wherein creating a second community if the target component in the first community among the plurality of communities satisfies the first preset condition comprises: creating a second community if a first target feature of the target component satisfies the first preset condition, wherein the first target feature is at least one feature specified by the user among a plurality of features of the target component. 7. The method according to claim 6, wherein the first target feature comprises: at least one of processor CPU utilization, memory utilization, and I / O throughput. 8. The method according to claim 7, wherein when the first target feature includes the CPU utilization rate, the first preset condition includes: the CPU utilization rate is greater than a preset CPU utilization rate; when the first target feature includes the memory utilization rate, the first preset condition includes: the memory utilization rate is greater than a preset memory utilization rate; when the first target feature includes the I / O throughput, the first preset condition includes: the I / O throughput is greater than a preset I / O throughput. 9.10. The method according to any one of claims 1-8, characterized in that, the method of creating the second community further includes: obtaining the modularity of the first community and each of the other communities in the plurality of communities excluding the first community, the modularity being a value used to describe the tightness of the calling relationship between communities; determining whether a target community exists in the other communities based on the modularity; the target community being a community in the other communities whose modularity with the first community is greater than a threshold; and creating the second community when the target community does not exist in the other communities. 10. The method according to claim 9, characterized in that, the method further includes: creating a second community combination when the target community exists in the other communities; the second community combination is the same as the first community combination including the first community and the target community, the second community combination including the second community and a third community, the third community being the same as the target community, and the second community combination being used to share part of the load of the first community combination. 11. The method according to any one of claims 1-10, characterized in that the method further comprises: 16 5 10 15 20 25 30 35 40 45 WO 2024 / 227365 PCT / CN2024 / 071300 If the target component does not meet the first preset condition, then determine hotspots and hot links according to the calling relationship between cloud service components in the first community; the hotspot is a cloud service component with a high calling frequency among multiple cloud service components in the first community, and the hot link is a link with a high calling frequency among the multiple cloud service components; process the hotspot and the hot link according to a preset reduction strategy; the preset reduction strategy is used to make the processed hotspot and the hot link meet their respective current load requirements. 12. The method according to claim 11, wherein the hotspot is the top N cloud service components with the highest popularity value among the multiple cloud service components included in the first community; wherein the popularity value is the sum of the out-degree and in-degree of the cloud service component, and N is an integer greater than 0; the hot link is the top m links with the highest call frequency among the multiple links between the multiple cloud service components included in the first community; m is an integer greater than 0. 13. The method according to claim 11 or 12, wherein processing the hotspot according to a preset strategy includes: determining a target hotspot reduction strategy from a first target correspondence based on the identifier of the community where the hotspot is located and the second target feature of the hotspot; the first target correspondence includes the identifiers of multiple communities, the correspondence between multiple third preset conditions and multiple hotspot reduction strategies; the second target feature is a feature pre-specified by the user among multiple features of the hotspot; the third preset condition is used to determine whether to expand the capacity of the hotspot;14. The method according to claim 13, wherein the hotspot reduction strategy includes: resetting expansion, and / or horizontal expansion. 15. The method according to any one of claims 11-14, wherein processing the hotspot link according to a preset strategy includes: determining a target hotspot link reduction strategy from a second target correspondence based on a third target feature of the hotspot link; the second target correspondence includes correspondences between multiple fourth preset conditions and multiple hotspot link reduction strategies; the third target feature is a feature pre-specified by the user among multiple features of the hotspot link; processing the hotspot link according to the target hotspot link reduction strategy. 16. The method according to claim 15, wherein the hotspot link reduction strategy includes: horizontal expansion, and / or vertical expansion. 17. The method according to claim 11 or 12, characterized in that, processing the hotspot and hot link according to a preset strategy comprises: determining a target type template from a plurality of preset type templates based on the distribution characteristics of the hotspot and the hot link; the distribution characteristics being used to characterize the spatial location information of the hotspot and the hot link; the target type template being the type template with the highest matching degree with the distribution characteristics among the plurality of preset type templates; determining a target reduction strategy corresponding to the target type template from the correspondence between the plurality of type templates and the plurality of reduction strategies; and processing the hotspot and the hot link according to the target reduction strategy. 18. A management device, characterized in that the management device comprises: a partitioning module and a creation module; the partitioning module is configured to partition the multiple cloud service components into multiple communities based on the calling relationships between multiple cloud service components in a cloud platform; wherein each community includes at least two cloud service components from the multiple cloud service components, the closeness of the calling relationships between the cloud service components included in each community is greater than or equal to a preset closeness, and the closeness between a cloud service component in any one of the multiple communities and a cloud service component in other communities is less than the preset closeness; the creation module is configured to create a second community if a target component in a first community of the multiple communities meets a first preset condition, the second community being the same as the first community, the second community being used to share part of the load of the first community; the first preset condition is used to indicate that the performance of the target component is lower than a preset performance, the first community being any one of the multiple communities; the target component is at least one of the multiple cloud service components in the first community. 19. The management device according to claim 18, wherein the invocation relationship includes one or more of the following: invocation information of multiple cloud service components in the cloud platform, any two cloud service components in the multiple cloud service components in the cloud platform.The frequency of calls between service components, and the out-degree and in-degree of any one of the multiple cloud service components in the cloud platform. 20. The management device according to claim 18 or 19, wherein the partitioning module is used to partition the multiple cloud service components into multiple communities based on the community discovery algorithm and the call relationship. 21. The management device according to any one of claims 18-20, wherein the target component includes: a cloud service component in the first community that meets a second preset condition, or at least one cloud service component specified by the user in the first community. 22. The management device according to claim 21, wherein the second preset condition includes: the top X cloud service components with the highest popularity value among the plurality of cloud service components in the first community; the popularity value is the sum of the out-degree and in-degree of the cloud service component, and X is an integer greater than 0. 23. The management device according to any one of claims 18-22, wherein the creation module is used to create the second community if the first target feature of the target component satisfies the first preset condition, wherein the first target feature is at least one feature specified by the user among the plurality of features of the target component. 24. The management device according to claim 23, wherein the first target feature includes: at least one of processor CPU utilization, memory utilization, and I / O throughput. 25. The management device according to claim 24, wherein when the first target feature includes the CPU utilization rate, the first preset condition includes: the CPU utilization rate is greater than a preset CPU utilization rate; when the first target feature includes the memory utilization rate, the first preset condition includes: the memory utilization rate is greater than a preset memory utilization rate; when the first target feature includes the I / O throughput, the first preset condition includes: the I / O throughput is greater than a preset I / O throughput. 26. The management device according to any one of claims 18-25, wherein the management device further includes: an acquisition module and a determination module; the acquisition module is used to acquire the module degree of the first community and each of the other communities in the plurality of communities other than the first community, the module degree being a value used to describe the tightness of the call relationship between communities; the determination module is used to determine, based on the module degree, whether a target community exists in the other communities; the target community is a community among the other communities whose module degree with the first community is greater than a threshold;The creation module is used to create a second community when the target community does not exist in the other communities. 27. The management device according to claim 26, wherein the creation module is used to create a second community combination when the target community exists in the other communities; the second community combination is the same as the first community combination including the first community and the target community, the second community combination includes the second community and a third community, the third community is the same as the target community, and the second community combination is used to share part of the load of the first community combination. 28. The management device according to any one of claims 18-27, wherein the determining module is used to determine hotspots and hot links according to the calling relationship between cloud service components in the first community if the target component does not meet the first preset condition; the hotspot is a cloud service component with a high calling frequency among multiple cloud service components in the first community, and the hot link is a link with a high calling frequency among the multiple cloud service components; the determining module is further used to process the hotspots and the hot links according to a preset reduction strategy; the preset reduction strategy is used to make the processed hotspots and the hot links meet their respective current load requirements. 29. The management device according to claim 28, wherein the hotspots are the top N cloud service components with the highest popularity values ​​among the multiple cloud service components included in the first community; wherein the popularity value is the sum of the out-degree and in-degree of the cloud service component, and N is an integer greater than 0; the hot links are the top m links with the highest call frequency among the multiple links between the multiple cloud service components included in the first community; m is an integer greater than 0. 30. The management device according to claim 27 or 28, wherein the determining module is configured to determine a target hotspot reduction strategy from a first target correspondence relationship based on the identifier of the community where the hotspot is located and the second target feature of the hotspot; the first target correspondence relationship includes the correspondence relationship between the identifiers of multiple communities, multiple third preset conditions and multiple hotspot reduction strategies; the second target feature is a feature pre-specified by the user among multiple features of the hotspot; the third preset condition is used to determine whether to expand the capacity of the hotspot; the determining module is further configured to process the hotspot according to the target hotspot reduction strategy. 31. The management device according to claim 30, wherein the hotspot reduction strategy includes: reset expansion, and / or horizontal expansion. 32. The management device according to any one of claims 28-31, wherein the determining module is configured to determine a target hotspot reduction strategy from a second target correspondence relationship based on the third target feature of the hotspot.The second target correspondence includes a correspondence between multiple fourth preset conditions and multiple hot link reduction strategies; the third target feature is a feature pre-specified by the user among multiple features of the hot link; the determining module is further configured to process the hot link according to the target hot link reduction strategy. 33. The management device according to claim 32, wherein the hot link reduction strategy includes: horizontal expansion and / or vertical expansion. 34. The management device according to claim 28 or 29, wherein the determining module is configured to determine a target type template from multiple preset type templates according to the distribution characteristics of the hot spot and the hot link; the distribution characteristics are used to characterize the location information of the hot spot and the hot link in space; the target type template is the type template with the highest matching degree with the distribution characteristics among the multiple preset type templates; the determining module is further configured to determine the target reduction strategy corresponding to the target type template from the correspondence between multiple type templates and multiple reduction strategies; and process the hot spot and the hot link according to the target reduction strategy. 35. A computing device cluster, characterized in that it comprises at least one computing device, each computing device including a processor and a memory; the processor of the at least one computing device is configured to execute instructions stored in the memory of the at least one computing device to cause the computing device cluster to perform the method as described in any one of claims 1 to 17. 36. A computer-readable storage medium, characterized in that it stores computer instructions, which, when executed on a computing device, cause the computing device to perform the method as described in any one of claims 1 to 17. 37. A computer program product comprising instructions, characterized in that, when executed by a computing device cluster, the instructions cause the computing device cluster to perform the method as described in any one of claims 1 to 17. 19 WO 2024 / 227365 PCT / CN2024 / 071300 Manual Drawings Cloud Platform System Platform Cloud Service A Cloud Service B Cloud Service N Service Component Service Component B 2 Service Component Service Fiber Service Component Service Component Management Equipment Figure 1 1 / 9 WO 2024 / 227365 PCT / CN2024 / 071300 Figure 2 2 / 9 WO 2024 / 227365 PCT / CN2024 / 071300 : 5: X - Second Community Q represents a cloud service component Figure 4 3 / 9 WO 2024 / 227365 PCT / CN2024 / 071300 Figure 6 4 / 9 WO 2024 / 227365 PCT / CN2024 / 071300 rij ( x This means that node i calls node j x times. rl 1(2)rl2(l) rl3(2) rl4(2) rl5(2) r21(l)r22(l)r23(5)r24(3)r25(8) r31(3)r32(4)r33(0) r34(2) r35(6) r41(2)r42(l)r43(3)r44(0)r45(2) r51(7)r52(2) r53(2) r54(9) r55(5) 5 / 9 WO 2024 / 227365 PCT / CN2024 / 071300 S210. The management device determines hotspots and hot links based on the call relationships between cloud service components in the first community. S220. The management device processes hotspots and hot links according to a preset mitigation strategy. S220a. The management device determines a target hotspot mitigation strategy from the first target correspondence based on the identifier of the community where the hotspot is located and the second target characteristic of the hotspot. S220b. The management device processes the hotspot according to the target hotspot mitigation strategy. S220c. The management device determines a target hotspot mitigation strategy from the second target correspondence based on the third target characteristic of the hotspot. S220d. The management device processes the hotspot according to the target hotspot mitigation strategy. Figure 8 6 / 9 WO 2024 / 227365 PCT / CN2024 / 071300 7 / 9 WO 2024 / 227365 PCT / CN2024 / 071300 Management Device Division Module 1101 1 I 1 Acquisition Module 1103 I --------------------L 1 | 1 Determination Module 1104 1 ___________________ L Create module 1102 Figure 11 Figure 12 8 / 9 WO 2024 / 227365 PCT / CN2024 / 071300 9 / 9 INTERNATIONAL SEARCH REPORT International application No. PCT / CN2024 / 071300 A. CLASSIFICATION OF SUBJECT MATTER H04L 67 / 1008(2022.01)1 According to International Patent Classification (IPC) or to both national classification and IPC B. FIELDS SEARCHED Minimum documentation searched (classification system followed by classification symbols) IPC:H04LDocumentation searched other than minimum documentation to the extent that such documents are included in the fields searched Electronic data base consulted during the international search (name of data base and, where practicable, search terms used) CNTXT, CNKI, ENTXT, ENTXTC, VEN: cloud, service component, dependency call relationship, cloning, load burden, load, cloud, service, component, dependen+, call+, clon+, load C. DOCUMENTS CONSIDERED TO BE RELEVANT Category* Citation of document, with indication, where appropriate, of the relevant passages Relevant to claim No. Y CN 112148484 A (PEKING UNIVERSITY) 29 December 2020 (2020-12-29) description, paragraphs 0001-0070 1-37 Y CN 109643249 A (ALCATEL LUCENT) 16 April 2019 (2019-04-16) description, paragraphs 0002-0092, and claims 1-15 1-37 A US 2021084108 A1 (HUAWEI TECHNOLOGIES CO., LTD.) 18 March 2021 (2021-03-18) entire document 1-37 | | Further documents are listed in the continuation of Box C. | / 1 See patent family annex. * Special categories of cited documents: “T" later documentpublished after the international filing date or priority “A" document defining the general state of the art which is not considered date and not in conflict with the application but cited to understand the to be of paiticulai' relevance principle or theory underlying the invention "D” document cited by the applicant in the international application “χ” document of particular relevance; the claimed invention cannot be “E” eailier application or patent but published on or after the international considered novel or cannot be considered to involve an inventive step filing date when the document is taken alone “L" document which may thi'ow doubts on priority claim(s) or which is “Y" document of paiticulai' relevance; the claimed invention cannot be cited to establish the publication date of another citation or other considered to involve an inventive step when the document is special reason (as specified) combined with one or more other such documents, such combination "O” documentrefen'ing to an oral disclosure, use, exhibition or other being obvious to a person skilled in the art means document member of the same patent family "P” document published prior to the international filing date but later than the priority date claimed Date of the actual completion of the international search 18 March 2024 Date of mailing of the international search report 22 March 2024 Name and mailing address of the ISA / CN China National Intellectual Property Administration (ISA / CN) China No. 6, Xitucheng Road, Jimenqiao, Haidian District, Beijing 100088 Authorized officer Telephone No. Form PCT / ISA / 210 (second sheet) (July 2022) International application No. PCT / CN2024 / 071300 INTERNATIONAL SEARCH REPORT Information on patent family members Patent document cited in search report Publication date (day / month / year) Patent family member(s) Publication date (day / month / year) CN 112148484 A 29 December 2020 None CN 109643249 A 16 April 2019 WO 2018029047 A2 15 February 2018 KR 20190039757A 15 April 2019 JP 2019525650 A 05 September 2019 EP 3282359 A1 14 February 2018 US 2021289385 A1 16 September 2021 US 2021084108 A1 18 March 2021 WO 2019228059 A1 05 December 2019 EP 4231151 A1 23 August 2023 EP 3790255 A10 March 2021 CN 108989384 A 11 December 2018 Form PCT / ISA / 210 (patent family annex) (July 2022) International Search Report International Application No. PCT / CN2024 / 071300 A. Subject Classification H04L 67 / 1008 (2022.01) 1. According to the International Patent Classification (IPC) or both the National Classification and IPC classifications. B. Minimum literature retrieved in the search field (indicate the classification system and classification number) IPC: H04L Electronic databases consulted during international searches (database name and search terms used, if applicable) CNTXT, CNKI, ENTXT, ENTXTC, VEN: cloud, service, component, dependency, call, clone, load, burden, load C. Relevant document types * Referenced documents, indicating relevant paragraphs where necessary Relevant claims Y CN 112148484 A (Peking University) December 29, 2020 (2020-12-29) Specification paragraphs 0001-0070 1-37 Y CN 109643249 A (Alcatel-Lucent) April 16, 2019 (2019-04-16) Specification paragraphs 0002-0092, claims 1-15 1-37 A US 2021084108 A1 (HUAWEI TECHNOLOGIES CO., LTD.) March 18, 2021 (2021-03-18) Full text 1-37 □ The remaining documents are listed on the continuation page in column C.*Specific types of cited documents: “A” Documents that are considered not particularly relevant and represent the general state of the prior art; “D” Documents cited by the applicant in an international application; “E” Prior applications or patents published on or after the international filing date; “L” Documents that may raise doubt about the priority claim, or documents cited to determine the publication date of another cited document, or documents cited for other specific reasons (as specifically stated); Documents involving disclosure, use, exhibition, or other forms of disclosure; “P” Documents whose publication date precedes the international filing date but later than the claimed priority date (see Annex to the patent family). "T is published after the application date or priority date, and does not conflict with the application, but is considered in isolation from document "X" for the purpose of understanding the inventive theory or principle, and the claimed invention is not novel or lacks inventiveness. Documents particularly related to "Y" are also considered, and the claimed invention is deemed not novel or lacking inventiveness. The claimed invention lacks inventiveness when the document is not combined with any of the other three patents of the same class and such combination is obviously incongruous with those skilled in the art. International search of patent family documents was actually completed on March 18, 2024. Name and mailing address of ISA / CN: China National Intellectual Property Administration, No. 6, Tucheng Road, Xijimenqiao West, Haidian District, Beijing, 100088, China. International search report mailing date: March 22, 2024. Authorized Officer: He Xijia. Telephone number: (+86) 010-53961586. PCT / ISA / 210 Form (Page 2) (July 2022)" International Search Report Information on Patent Families International Application No. PCT / CN2024 / 071300 Publication Dates of Patent Documents Cited in the Search Report (Year / Month / Day) Publication Dates of Patent Families (Year / Month / Day) CN 112148484 A December 29, 2020 None CN 109643249 A April 16, 2019 WO 2018029047 A2 February 15, 2018 KR 20190039757 A April 15, 2019 JP 2019525650 A September 5, 2019 EP 3282359 A1 February 14, 2018 US 2021289385 A1 September 16, 2021 US 2021084108 A1 March 18, 2021 WO 2019228059 Al December 5, 2019 EP 4231151 Al August 23, 2023 EP 3790255 Al March 10, 2021 CN 108989384 A December 11, 2018 PCT / ISA / 210 Form (Appendix to Patent Family) (July 2022) (19) *EP004697678A1* (11) EP 4 697 678 A1 (12)EUROPEAN PATENT APPLICATION published in accordance with Art. 153(4) EPC (43) Date of publication: 18.02.2026 Bulletin 2026 / 08 (21) Application number: 24799823.0 (22) Date of filing: 09.01.2024 (51) International Patent Classification (IPC): H04L 67 / 1008 (2022.01) (52) Cooperative Patent Classification (CPC): H04L 67 / 1008 (86) International application number: PCT / CN2024 / 071300 (87) International publication number: WO 2024 / 227365 (07.11.2024 Gazette 2024 / 45) (84) Designated Contracting States: AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC ME MK MT NL NO PL PT RO RS SE SI SK SM TR Designated Extension States: BA Designated Validation States: KH MA MD TN (30) Priority: 04.05.2023 CN 202310492451 05.07.2023 CN 202310820305 (71) Applicant: Huawei Cloud Computing Technologies Co., Ltd. Guiyang, Guizhou 550025 (CN) (72) Inventors: • ZHANG, Lin Guiyang, Guizhou 550025 (CN) • DUAN, Bin Guiyang, Guizhou 550025 (CN) • WU, Zhou Guiyang, Guizhou 550025 (CN) (74)Representative: Isarpatent Patent‑ und Rechtsanwälte Partg mbB Friedrichstraße 31 80801 München (DE) (54) CLOUD PLATFORM MANAGEMENT METHOD AND DEVICE, PROGRAM PRODUCT, AND STORAGE MEDIUM (57) Embodiments of this application relate to the field of cloud computing, and provide a cloud platform management method and an apparatus, a program pro- duct, and a storage medium, to quickly adjust load pres- sure of a cloud platform. The method includes: grouping a plurality of cloud service components in a cloud platform into a plurality of communities based on call relationships between the plurality of cloud service components, where a degree of closeness of a call relationship be- tween cloud service components included in each com- munity is greater than or equal to a preset degree of closeness, and a degree of closeness between a cloud service component in any one of the plurality of commu- nities and a cloud service component in another com- munity is less than the preset degree ofcloseness; and if a target component in a first community of the plurality of communities meets a first preset condition, creating a second community, where the second community is the same as the first community, the second community is configured to share a part of load of the first community, the first preset condition indicates that performance of the target component is lower than preset performance, and the target component is at least one of a plurality of cloud service components in the first community. EP 4 69 7 67 8 A 1 Processed by Luminess, 75001 PARIS (FR) (Cont. next page) 2 EP 4 697 678 A1 Description

[0001] This application claims priority to Chinese Patent Application No. 202310492451.0, filed with the China National Intellectual Property Administration on May 4, 2023 and entitled "SERVICE PERFORMANCE ANALYSIS METHOD AND SYSTEM", and Chinese Patent Application No. 202310820305.6, filed with the China National Intellectual Property Administration on July 5, 2023, andentitled "CLOUD PLATFORM MANAGEMENT METHOD AND APPARATUS, PROGRAM PRODUCT, AND STORAGE MEDIUM", both of which are incorporated herein by reference in their entireties. TECHNICAL FIELD

[0002] Embodiments of this application relate to the field of cloud computing, and in particular, to a cloud platform management method and an apparatus, a program product, and a storage medium. BACKGROUND

[0003] Cloud services (Cloud services) are required services obtained on demand in a scalable manner through a network, based on a model for using, and delivering related services over the Internet. A cloud service means that a computing capability can also be circulated through the Internet as a commodity. A plurality of cloud services constitute a cloud platform, and the cloud platform is configured to provide a powerful computing service for a user. Each cloud service may include a plurality of cloud service components, and dependency relationships and / or association relationships exist between theplurality of cloud service components.

[0004] As application scenarios of cloud services increase continuously, customers pay more attention to processing capabilities of cloud platforms or cloud services under heavy load. A common method for processing pressure on cloud service components in a cloud platform includes: operation and maintenance personnel monitor an out-degree and an in- degree of each cloud service component in the cloud platform and a data transmission frequency of a link between the cloud service components, and determine, based on experience, whether any hot point and hot link exist in the cloud platform. When a hot point and a hot link exist in the cloud platform, the operation and maintenance personnel perform capacity expansion on the hot point and the hot link based on experience.

[0005] However, during the capacity expansion of the hot point or the hot link, the operation and maintenance personnel can only adjust load of cloud service components one by one, andcannot quickly adjust load pressure of the cloud platform, resulting in limited performance of the cloud platform. SUMMARY

[0006] Embodiments of this application provide a cloud platform management method and an apparatus, a program product, and a storage medium, to quickly adjust load pressure of a cloud platform.

[0007] To achieve the foregoing objective, the following technical solutions are used in embodiments of this application: According to a first aspect, an embodiment of this application provides a cloud platform management method. The method includes: grouping a plurality of cloud service components in a cloud platform into a plurality of communities based on call relationships between the plurality of cloud service components, where each community includes at least two of the plurality of cloud service components, a degree of closeness of a call relationship between cloud service components included in each community is greater than or equal to a preset degree of closeness,and a degree of closeness between a cloud service component in any one of the plurality of communities and a cloud service component in another one of the plurality of communities is less than the preset degree of closeness; and if a target component in a first community of the plurality of communities meets a first preset condition, creating a second community, where the second community is the same as the first community, the second community is configured to share a part of load of the first community, the first preset condition indicates that performance of the target component is lower than preset performance, the first community is any one of the plurality of communities, and the target component is at least one of a plurality of cloud service components in the first community.

[0008] Based on this, an embodiment of this application provides a cloud platform management method. The method includes: grouping, based on call relationships between a plurality of cloud servicecomponents in a cloud platform, the plurality of cloud service components into a plurality of communities including a first community, where a degree of closeness of a call relationship between cloud service components in a community is relatively high, and a degree of closeness of a call relationship between communities is relatively low; and when the first community meets a preset condition, creating a second community same as the first community (for example, cloning the first community to obtain the second community), so that the second community shares a part of load of the first community. Therefore, when load pressure of the cloud platform is relatively high, an entire community including a plurality of cloud service components is cloned. Because a degree of closeness of a call relationship between cloud service components in the entire community is 3 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 relatively high, load balancing can be performed on the first community and thesecond community to quickly adjust the load pressure of the cloud platform, thereby improving working performance of the cloud platform.

[0009] In a possible implementation, the call relationship includes one or more of the following: call information of the plurality of cloud service components in the cloud platform, a call frequency of a link between any two of the plurality of cloud service components in the cloud platform, and an out-degree and an in-degree of any one of the plurality of cloud service components in the cloud platform.

[0010] In a possible implementation, the call information includes: an identifier of a cloud service component called by a first component when the first component is a calling component; and an identifier of a cloud service component calling the first component when the first component is a called component, where the first component is any one of the plurality of cloud service components in the cloud platform.

[0011] In a possible implementation,grouping the plurality of cloud service components in the cloud platform into the plurality of communities based on the call relationships between the plurality of cloud service components includes: grouping the plurality of cloud service components into the plurality of communities based on a community discovery algorithm and the call relationships.

[0012] In a possible implementation, the target component includes: a cloud service component that meets a second preset condition among the plurality of cloud service components in the first community, or at least one cloud service component specified by a user among the plurality of cloud service components in the first community.

[0013] In a possible implementation, the second preset condition includes top X cloud service components with highest popularity values among the plurality of cloud service components in the first community, where the popularity value is a sum of an out-degree of a cloud service component and an in-degree ofthe cloud service component, and X is an integer greater than 0.

[0014] In a possible implementation, creating the second community if the target component in the first community of the plurality of communities meets the first preset condition includes: if a first target feature of the target component meets the first preset condition, creating the second community, where the first target feature is at least one feature specified by the user among a plurality of features of the target component.

[0015] In a possible implementation, the first target feature includes at least one of processor CPU usage, memory usage, and an I / O throughput.

[0016] In a possible implementation, when the first target feature includes the CPU usage, the first preset condition includes: the CPU usage is higher than preset CPU usage; when the first target feature includes the memory usage, the first preset condition includes: the memory usage is higher than preset memory usage; or when the first target featureincludes the I / O throughput, the first preset condition includes: the I / O throughput is higher than a preset I / O throughput.

[0017] In a possible implementation, creating the second community includes: obtaining modularity between the first community and each of other communities than the first community among the plurality of communities, where the modularity is a value used to describe a degree of closeness of a call relationship between communities; determining, based on the modularity, whether a target community exists among the other communities, where the target community is a community whose modularity with the first community is greater than a threshold among the other communities; and when the target community does not exist among the other communities, creating the second community.

[0018] In a possible implementation, when the target community exists among the other communities, a second community combination is created, where the second community combination is the same asa first community combination including the first community and the target community, the second community combination includes the second community and a third community, the third community is the same as the target community, and the second community combination is used to share a part of load of the first community combination.

[0019] In this embodiment of this application, it is determined whether the target community whose modularity with the first community is greater than the threshold exists among the plurality of communities; and when the target community exists (that is, a community whose degree of closeness of a call relationship with the first community is relatively high exists), the first community and the target community are used as an entire community (that is, the first community combination), and the second community combination that includes the second community and the third community and that is the same as the first community combination is created, where thesecond community combination is used to share load of the first community combination, so as to resolve a problem that load pressure of the target community is excessively high because capacity expansion is performed only on the first community, and further improve effectiveness of reducing the load pressure of the cloud platform.

[0020] In a possible implementation, the method further includes: if the target component does not meet the first preset condition, determining a hot point and a hot link based on a call relationship between cloud service components in the first community, where the hot point is a cloud service component with a relatively high call frequency among the plurality of cloud service components in the first community, and the hot link is a link with a relatively high call frequency between the plurality of cloud service components; and processing the hot point and the hot link based on a preset mitigation policy, where the preset mitigation policy is used to enablethe processed hot point and hot link to meet respective current load requirements. 4 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55

[0021] In the foregoing embodiment, when the target component in the first community does not meet the first preset condition, the hot point and the hot link in the first community are determined; and then a target hot point mitigation policy is determined based on the community in which the hot point is located and a third preset condition that a second target feature of the hot point meets, and the hot point is processed based on the target hot point mitigation policy, so that load pressure of the hot point is mitigated. Similarly, a management apparatus determines that a hot link mitigation policy corresponding to a fourth preset condition that a third target feature of the hot link meets is a target hot link mitigation policy, and processes the hot link based on the target hot link mitigation policy, to reduce load pressure of the processed hot link,thereby resolving a problem that the cloud platform breaks down due to relatively heavy load of at least one cloud service component and link in the first community.

[0022] In a possible implementation, the hot point is top N cloud service components with highest popularity values among the plurality of cloud service components included in the first community, where the popularity value is a sum of an out-degree of a cloud service component and an in-degree of the cloud service component, and N is an integer greater than 0; and the hot link is top M links with highest call frequencies among a plurality of links between the plurality of cloud service components included in the first community, where M is an integer greater than 0.

[0023] In a possible implementation, processing the hot point based on the preset policy includes: determining a target hot point mitigation policy from a first target correspondence based on an identifier of a community in which the hot point is located and asecond target feature of the hot point, where the first target correspondence includes a correspondence between identifiers of the plurality of communities, a plurality of third preset conditions, and a plurality of hot point mitigation policies, the second target feature is a feature specified in advance by the user among a plurality of features of the hot point, and the third preset condition is used to determine whether to perform capacity expansion on the hot point; and processing the hot point based on the target hot point mitigation policy.

[0024] In a possible implementation, the hot point mitigation policy includes: reset scaling and / or horizontal scaling.

[0025] In a possible implementation, processing the hot link based on the preset policy includes: determining a target hot link mitigation policy from a second target correspondence based on a third target feature of the hot link, where the second target correspondence includes a correspondence between a plurality of fourthpreset conditions and a plurality of hot link mitigation policies, and the third target feature is a feature specified in advance by the user among a plurality of features of the hot link; and processing the hot link based on the target hot link mitigation policy.

[0026] In a possible implementation, the hot link mitigation policy includes: lateral scaling and / or longitudinal scaling.

[0027] In a possible implementation, processing the hot point and the hot link based on the preset policy includes: determining a target type template from a plurality of preset type templates based on a distribution feature of the hot point and the hot link, where the distribution feature is used to represent spatial location information of the hot point and the hot link, and the target type template is a type template having a highest degree of matching with the distribution feature among the plurality of preset type templates; determining, from a correspondence between a plurality of type templates anda plurality of mitigation policies, a target mitigation policy corresponding to the target type template; and processing the hot point and the hot link based on the target mitigation policy.

[0028] In the foregoing embodiment, when the target component in the first community does not meet the first preset condition, the hot point and the hot link in the first community are determined; and then the management apparatus determines the target type template from the plurality of preset type templates based on the distribution feature of the hot point and the hot link, and processes the hot point and the hot link based on the target mitigation policy corresponding to the target type template, to reduce load pressure of the processed hot point and hot link, thereby resolving a problem that the cloud platform breaks down due to relatively heavy load of at least one cloud service component and link in the first community.

[0029] According to a second aspect, this application provides amanagement apparatus. The management apparatus includes a grouping module and a creation module. The grouping module is configured to group a plurality of cloud service components in a cloud platform into a plurality of communities based on call relationships between the plurality of cloud service components, where each community includes at least two of the plurality of cloud service components, a degree of closeness of acall relationship betweencloud service components included in eachcommunity is greater than orequal to a preset degree of closeness, and a degree of closeness between a cloud service component in any one of the plurality of communities and a cloud service component in another one of the plurality of communities is less than the preset degree of closeness; and the creation module is configured to create a second community if a target component in a first community of the plurality of communities meets a first preset condition, where the second community is the same asthe first community, the second community is configured to share a part of load of the first community, the first preset condition indicates that performance of the target component is lower than preset performance, the first community is any one of the plurality of communities, and the target component is at least one of a plurality of cloud service components in the first community.

[0030] In a possible implementation, the call relationship includes one or more of the following: call information of the plurality of cloud service components in the cloud platform, a call frequency of a link between any two of the plurality of cloud service components in the cloud platform, and an out-degree and an in-degree of any one of the plurality of cloud 5 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 service components in the cloud platform, where the call information includes: an identifier of a cloud service component called by a first component when the first component is a callingcomponent; and an identifier of a cloud service component calling the first component when the first component is a called component, where the first component is any one of the plurality of cloud service components in the cloud platform.

[0031] In a possible implementation, the grouping module is specifically configured to group the plurality of cloud service components into the plurality of communities based on a community discovery algorithm and the call relationships.

[0032] In a possible implementation, the target component includes: a cloud service component that meets a second preset condition among the plurality of cloud service components in the first community, or at least one cloud service component specified by a user among the plurality of cloud service components in the first community.

[0033] In a possible implementation, the second preset condition includes top X cloud service components with highest popularity values among the plurality of cloud service components inthe first community, where the popularity value is a sum of an out-degree of a cloud service component and an in-degree of the cloud service component, and X is an integer greater than 0.

[0034] In a possible implementation, the creation module is configured to create the second community if a first target feature of the target component meets the first preset condition, where the first target feature is at least one feature specified by the user among a plurality of features of the target component.

[0035] In a possible implementation, the first target feature includes at least one of processor CPU usage, memory usage, and an I / O throughput.

[0036] In a possible implementation, when the first target feature includes the CPU usage, the first preset condition includes: the CPU usage is higher than preset CPU usage; when the first target feature includes the memory usage, the first preset condition includes: the memory usage is higher than preset memory usage; or when the first targetfeature includes the I / O throughput, the first preset condition includes: the I / O throughput is higher than a preset I / O throughput.

[0037] In a possible implementation, the management apparatus further includes an obtaining module and a determin- ing module, where the obtaining module is configured to obtain modularity between the first community and each of other communities than the first community among the plurality of communities, where the modularity is a value used to describe a degree of closeness of a call relationship between communities; the determining module is configured to determine, based on the modularity, whether a target community exists among the other communities, where the target community is a community whose modularity with the first community is greater than a threshold among the other communities; and the creation module is configured to create the second community when the target community does not exist among the other communities.

[0038] In a possibleimplementation, the creation module is configured to create a second community combination when the target community exists among the other communities, where the second community combination is the same as a first community combination including the first community and the target community, the second community combination includes the second community and a third community, the third community is the same as the target community, and the second community combination is used to share a part of load of the first community combination.

[0039] In a possible implementation, the determining module is configured to determine a hot point and a hot link based on a call relationship between cloud service components in the first community if the target component does not meet the first preset condition, where the hot point is a cloud service component with a relatively high call frequency among the plurality of cloud service components in the first community, and the hot link is a link with arelatively high call frequency between the plurality of cloud service components; and the determining module is further configured to process the hot point and the hot link based on a preset mitigation policy, where the preset mitigation policy is used to enable the processed hot point and hot link to meet respective current load requirements.

[0040] In a possible implementation, the hot point is top N cloud service components with highest popularity values among the plurality of cloud service components included in the first community, where the popularity value is a sum of an out-degree of a cloud service component and an in-degree of the cloud service component, and N is an integer greater than 0; and the hot link is top M links with highest call frequencies among a plurality of links between the plurality of cloud service components included in the first community, where M is an integer greater than 0.

[0041] In a possible implementation, the determining module is configured todetermine a target hot point mitigation policy from a first target correspondence based on an identifier of a community in which the hot point is located and a second target feature of the hot point, where the first target correspondence includes a correspondence between identifiers of the plurality of communities, a plurality of third preset conditions, and a plurality of hot point mitigation policies, the second target feature is a feature specified in advance by the user among a plurality of features of the hot point, and the third preset condition is used to determine whether to perform capacity expansion on the hot point; and the determining module is further configured to process the hot point based on the target hot point mitigation policy.

[0042] In a possible implementation, the hot point mitigation policy includes: reset scaling and / or horizontal scaling.

[0043] In a possible implementation, the determining module is configured to determine a target hot link mitigation policyfrom a second target correspondence based on a third target feature of the hot link, where the second target correspon- dence includes a correspondence between a plurality of fourth preset conditions and a plurality of hot link mitigation 6 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 policies, and the third target feature is a feature specified in advance by the user among a plurality of features of the hot link; and the determining module is further configured to process the hot link based on the target hot link mitigation policy.

[0044] In a possible implementation, the hot link mitigation policy includes: lateral scaling and / or longitudinal scaling.

[0045] In a possible implementation, the determining module is configured to determine a target type template from a plurality of preset type templates based on a distribution feature of the hot point and the hot link, where the distribution feature is used to represent spatial location information of the hot point and the hot link,and the target type template is a type template having a highest degree of matching with the distribution feature among the plurality of preset type templates; and the determining module is further configured to determine, from a correspondence between a plurality of type templates and a plurality of mitigation policies, a target mitigation policy corresponding to the target type template; and process the hot point and the hot link based on the target mitigation policy.

[0046] According to a third aspect, this application provides a computing device cluster, including at least one computing device, where each computing device includes a processor and a memory; and the processor of the at least one computing device is configured to execute instructions stored in the memory of the at least one computing device, so that the computing device cluster performs the method according to any one of the first aspect and the possible implementations of the first aspect.

[0047] According to afourth aspect, this application provides a computer-readable storage medium. The computer- readable storage medium stores computer instructions. When the computer instructions are run on a computing device, the computing device is enabled to perform the method according to any one of the first aspect and the possible implementations of the first aspect.

[0048] According to a fifth aspect, this application provides acomputer program product including instructions. When the computer program product is run on a computer, the computer is enabled to perform the method according to any one of the first aspect and the possible implementations of the first aspect.

[0049] It should be understood that for beneficial effects achieved by technical solutions in the second aspect to the fifth aspect and the corresponding possible implementations in this application, refer to the foregoing technical effects in the first aspect and the corresponding possible implementations of the first aspect.Details are not described herein again. BRIEF DESCRIPTION OF DRAWINGS

[0050] FIG. 1 is a diagram of a cloud platform system according to an embodiment of this application; FIG. 2 is a first schematic flowchart of a cloud platform management method according to an embodiment of this application; FIG. 3 is a first diagram of a call relationship graph according to an embodiment of this application; FIG. 4 is a second diagram of a call relationship graph according to an embodiment of this application; FIG. 5 is a second schematic flowchart of a cloud platform management method according to an embodiment of this application; FIG. 6 is a third schematic flowchart of a cloud platform management method according to an embodiment of this application; FIG. 7 is a diagram for determining a hot point and a hot link according to an embodiment of this application; FIG. 8 is a fourth schematic flowchart of a cloud platform management method according to an embodiment of this application; FIG. 9 is afifth schematic flowchart of a cloud platform management method according to an embodiment of this application; FIG. 10 is a diagram of a type template according to an embodiment of this application; FIG. 11 is a diagram of a structure of a management apparatus according to an embodiment of this application; FIG. 12 is a diagram of a computing device cluster according to an embodiment of this application; and FIG. 13 is a network connection diagram according to an embodiment of this application. DESCRIPTION OF EMBODIMENTS

[0051] The term "and / or" in this specification merely describes an association relationship between associated objects and represents that three relationships may exist. For example, A and / or B may represent the following three cases: Only A exists, both A and B exist, and only B exists.

[0052] In the specification and claims in embodiments of this application, the terms "first", "second", and the like are intended to differentiate different objects, but do notdescribe a specific order of the objects. For example,a first community, a second community, and the like are used to differentiate different communities, and are not used to describe a particular order of the communities. 7 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55

[0053] In embodiments of this application, the term "example", "for example", or the like is used to represent an example, an illustration, or a description. Any embodiment or design scheme described as an "example" or "for example" in embodiments of this application should not be construed as being more preferred or advantageous than another embodiment or design scheme. Exactly, use of the term "example" or "for example" or the like is intended to present a relative concept in a specific manner.

[0054] In descriptions of embodiments of this application, unless otherwise stated, "a plurality of" means two or more. For example, a plurality of processing units are two or more processing units, and a plurality of systemsare two or more systems.

[0055] First, some concepts related to a cloud platform management method and an apparatus, a program product, and a storage medium provided in embodiments of this application are described.

[0056] Cloud service component: is also referred to as a cloud component or a component. One cloud service includes a plurality of cloud service components, and the plurality of cloud service components provide all services in the cloud service, that is, each cloud service component provides a sub-service for the cloud service. For example, it is assumed that a cloud service A is used to provide a data storage service, and that the cloud service A includes a data cache component and a data persistence component, where the data cache component is configured to provide a service for caching to-be- persisted data, and the data persistence component is configured to provide a service for persisting the cached data. In this case, the service provided by the data cache componentand the service provided by the data persistence component together constitute the cloud service A.

[0057] When the cloud service component is a MySQL component, the cloud service component is configured to provide a MySQL database service.

[0058] Out-degree and in-degree: In a directed graph, an arrow has a direction and points from one vertex to another vertex. In this way, a quantity of arrows pointing to each vertex is an in-degree of the vertex. A quantity of arrows pointing out from this vertex is an out-degree of the vertex. In other words, an in-degree of the cloud service component is a quantity of times of sending data to the cloud service component within a unit time, or a quantity of times of calling the cloud service component; and an out-degree of the cloud service component is a quantity of times of sending data by the cloud service component to another cloud service component within a unit time, or a quantity of times of calling another cloud service component.

[0059] Modularity (modularity): is also referred to as a modularity metric, and is a common method for measuring strength of a network community structure, that is, the modularity is a value for measuring a degree of closeness of a community.

[0060] Vertical scaling: means enhancing hardware performance of a single machine, for example, increasing a quantity of CPU cores, for example, to 32 cores, upgrading to a better network adapter such as 10 GE, upgrading to a better hard disk such as an SSD, expanding a hard disk capacity, for example, to 2 TB, and expanding a system memory, for example, to 128 GB.

[0061] Horizontal scaling: means adding more servers or program instances to distribute load, thereby improving storage and computing capabilities, for example, adding storage devices.

[0062] Lateral scaling: means adding more equivalent functional components in parallel to distribute load.

[0063] Longitudinal scaling: is a process of increasing a capacity of a given instance, that is,expanding a capacity of a point to process more requests.

[0064] As vendors pay more attention to processing capabilities of cloud platforms or cloud services under heavy load pressure, a common method for processing pressure on cloud service components in a cloud platform is provided, including: operation and maintenance personnel monitor an out-degree and an in-degree of each cloud service component in the cloud platform and a data transmission frequency of a link between the cloud service components, and determine, based on experience, whether any hot point and hot link exist in the cloud platform. When a hot point and a hot link exist in the cloud platform, the operation and maintenance personnel perform capacity expansion on the hot point and the hot link based on experience.

[0065] For example, a relationship graph including call relationships of a plurality of cloud service components includes a cloud service component A and a cloud service component B. The cloud servicecomponent B is configured to store data sent by the cloud service component A. When operation and maintenance personnel determine that the cloud service component A is a hot point, capacity expansion is performed on the cloud service component A, so that the cloud service component A can normally process service data under current load pressure.

[0066] However, when load pressure of the cloud platform is relatively high, after capacity expansion is performed only on the cloud service component A as a hot point, although processing efficiency of the cloud service component A is improved, load pressure of the cloud service component B is increased, and consequently, the cloud service component B becomes a new hot point. It can be learned that, in the foregoing method, when the load pressure of the cloud platform is relatively high, the load pressure of the cloud platform cannot be quickly adjusted, resulting in limited performance of the cloud platform.

[0067] Based on this, anembodiment of this application provides a cloud platform management method. The method 8 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 includes: grouping, based on call relationships between a plurality of cloud service components in a cloud platform, the plurality of cloud service components into a plurality of communities including a first community, where a degree of closeness of a call relationship between cloud service components in a community is relatively high, and a degree of closeness of a call relationship between communities is relatively low; and when the first community meets a preset condition, creating a second community same as the first community (for example, cloning the first community to obtain the second community), so that the second community shares a part of load of the first community. Therefore, when load pressure of the cloud platform is relatively high, an entire community including a plurality of cloud service components is cloned. Because a degree ofcloseness of a call relationship between cloud service components in the entire community is relatively high, load balancing can be performed on the first community and the second community to quickly adjust the load pressure of the cloud platform, thereby improving working performance of the cloud platform.

[0068] An embodiment of this application provides a cloud platform management method. The method may be applied to a cloud platform system shown in FIG. 1. The cloud platform system includes a cloud platform and a management apparatus.

[0069] The cloud platform is configured to provide a powerful computing service for a user. The cloud platform includes N cloud services, and a dependency relationship exists between the N cloud services. One of the N cloud services is used to provide a required service (for example, a storage service or a gateway service) for the cloud platform. Each cloud service includes X cloud service components, a dependency relationship may exist between theplurality of cloud service components, and one of the X cloud service components is configured to provide a sub-service for the cloud service.

[0070] It should be noted that any two of the N cloud services may include different quantities of cloud service components.

[0071] The management apparatus is configured to perform the cloud platform management method provided in embodiments of this application, to manage the cloud platform, so that when load pressure of the cloud platform is relatively high, the management apparatus quickly reduces load pressure of a cloud service in the cloud platform. The management apparatus performs the cloud platform management method provided in embodiments of this application in S110 to S 140 below. Details are not described herein again.

[0072] It should be understood that the management apparatus may be a cloud server in the cloud platform, or may be a cloud server independent of the cloud platform. Specifically, a specific location of the managementapparatus is not limited in this embodiment of this application.

[0073] It should be noted that a system architecture and an application scenario described in embodiments of this application are intended to describe the technical solutions in embodiments of this application more clearly, and do not constitute any limitation on the technical solutions provided in embodiments of this application. Persons of ordinary skill in the art may know that, with evolution of the system architecture and emergence of a new service scenario, the technical solutions provided in embodiments of this application are also applicable to a similar technical problem.

[0074] An embodiment of this application provides a cloud platform management method. The method is applied to the management apparatus in the cloud platform system shown in FIG. 1. As shown in FIG. 2, the method includes S110 to S140.

[0075] S110: A management apparatus obtains call relationships between a plurality of cloud service componentsin a cloud platform.

[0076] The cloud platform is a cloud platform including a plurality of cloud service components, as shown in FIG. 1, and mutual call relationships exist between the plurality of cloud service components.

[0077] The call relationship includes one or more of the following: call information of the plurality of cloud service components, a call frequency of a link between any two of the plurality of cloud service components, and an out-degree and an in-degree of each of the plurality of cloud service components.

[0078] The call information includes: an identifier of another cloud service component called by a first component when the first component is a calling component; and an identifier of a cloud service component calling the first component when the first component is a called component, where the first component is any one of the plurality of cloud service components. In other words, the call information includes a correspondence between a calling end and acalled end among the plurality of cloud service components. For example, when a cloud service component A calls a cloud service component B, the calling end is the cloud service component A, and the called end is the cloud service component B, that is, the cloud service component A→the cloud service component B.

[0079] The call frequency of the link indicates a quantity of times that the link transmits data within a unit time.

[0080] For example, the call relationship may be represented by using a call relationship graph shown in FIG. 3. Each node (for example,②) in the graph represents a cloud service component (for example, a cloud service component 2). Any node (for example, a node 2 representing the cloud service component 2) in the graph includes an identifier of another cloud service component called when the cloud service component 2 is used as a calling end, for example,②→①, and an identifier of a cloud service component of a calling end calling the cloud service component 2when the cloud service component 2 is used as a called end, for example,⑧→② and⑥→②; also includes a call frequency of a link between the cloud service component 2 and a cloud service component that has a call relationship with the cloud service component 2, 9 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 for example, a call frequency of a link 1 between (2) and⑥; and further includes an out-degree and an in-degree of the cloud service component 2.

[0081] It should be noted that a specific implementation process of S110 may be that the management apparatus collects the call relationship from the cloud platform by using a monitoring service (for example, a network management service), or may be that a user pre-imports the call relationship into the management apparatus and then the management apparatus directly obtains the call relationship locally. Specifically, a specific implementation of S110 is not limited in this embodiment of this application.

[0082] S120: The managementapparatus groups the plurality of cloud service components in the cloud platform into a plurality of communities based on the call relationships between the plurality of cloud service components.

[0083] A degree of closeness of a call relationship between cloud service components in any one of the plurality of communities is greater than or equal to a preset degree of closeness, and a degree of closeness between a cloud service component in any one of the plurality of communities and a cloud service component in another one of the plurality of communities is less than the preset degree of closeness. In other words, a degree of closeness of a call relationship between any two of the plurality of communities is less than the preset degree of closeness, where the degree of closeness of the call relationship is used to represent a call frequency between two cloud service components. In other words, among the plurality of cloud service components, cloud service components with a relativelyhigh degree of closeness of a call relationship are grouped into one community; and among the plurality of cloud service components, cloud service components with a relatively low degree of closeness of a call relationship are grouped into different communities.

[0084] A specific implementation of S120 may be that the management apparatus groups the plurality of cloud service components into the plurality of communities based on a community discovery algorithm and the call relationships between the plurality of cloud service components; or may be that the management apparatus groups the plurality of cloud service components into the plurality of communities based on a preset training model. Specifically, a specific imple- mentation of S120 is not limited in this embodiment of this application.

[0085] It should be noted that, when the foregoing S120 is that the management apparatus groups the plurality of cloud service components into the plurality of communities based on the communitydiscovery algorithm, a specific imple- mentation of S120 includes: using each of the plurality of cloud service components as an independent community; then calculating modularity Q of each of the plurality of communities based on the following formula (1); when it is necessary to determine whether to group a community A and a community B into one community, calculating modularity Q of a community C obtained by grouping the community A and the community B into one community; and when a difference between the modularity Q of the community C and modularity Q of the community A is greater than 0, grouping the community A and the community B into one community; or when a difference between the modularity Q of the community C and modularity Q of the community A is less than or equal to 0, not grouping the community A and the community B into one community. The process is repeated until a value of modularity Q of a community (for example, a community 3) obtained by merging any twocommunities (for example, a community 1 and a community 2) among the plurality of communities obtained through grouping is greater than modularity Q of the community 1 and modularity Q of the community 2.

[0086] In the formula, m is a sum of weights of all edges in a call network corresponding to the call relationship; Σ in is a sum of weights of all edges between nodes in the community A; and Σ tot is a sum of weights of all edges connected to nodes in the community A, that is, Σ tot is a sum of weights of all edges in the community A (weights of internal edges of the community A for short), and weights of all edges between nodes in other communities and the nodes in the community A (weights of external edges of the community A for short).

[0087] It should be noted that a quantity of edges between one node and another node indicates a quantity of calls between the one node and the another node within a unit time.

[0088] For example, it is assumed that the call network corresponding tothe call relationship is shown in FIG. 3. Based on the community discovery algorithm, the call network is divided into three communities.Specifically, as shown in FIG. 4, a node 6 to a node 10 are grouped into a first community, a node 1 to a node 5 are grouped into a second community, and a node 11 to a node 15 are grouped into a third community. A degree of closeness of an edge connection between a plurality of nodes in any one of the three communities is relatively high, and a degree of closeness of an edge connection between any two communities is relatively low.

[0089] S130: The management apparatus determines whether a target component in a first community meets a first preset condition.

[0090] It should be understood that the first community is any one of the plurality of communities, and that the first community includes a plurality of cloud service components. 10 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55

[0091] It should be noted that heavier load of the first communityindicates lower performance of the target component in the first community. Based on this, the first preset condition indicates that performance of the target component is lower than preset performance.

[0092] In an implementation, the target component may be a cloud service component that meets a second preset condition among the plurality of cloud service components in the first community, where the second preset condition includes top X cloud service components with highest popularity values among the plurality of cloud service components in the first community, or top M cloud service components with lowest popularity values among the plurality of cloud service components in the first community, where a popularity value of a cloud service component is a sum of an out-degree of the cloud service component and an in-degree of the cloud service component, and both X and M are integers greater than 0. In other words, the management apparatus calculates popularity values of all cloudservice components in the first community, and sorts the popularity values; and then determines that the top X cloud service components with the highest popularity values are target components, or determines that the top M cloud service components with the lowest popularity values are target components.

[0093] For example, it is assumed that the first community includes five cloud service components: a cloud service component A to a cloud service component E, and that the five cloud service components are sorted in descending order of popularity values as follows: cloud service component B, cloud service component C, cloud service component E, cloud service component D, and cloud service component A. It is also assumed that the second preset condition includes top two cloud service components with the highest popularity values among the plurality of cloud service components in the first community. In this case, the target components are the cloud service component B and the cloudservice component C.

[0094] In another implementation, the target component is at least one cloud service component specified by the user among the plurality of cloud service components in the first community. In other words, the target component is specified by the user in advance in the first community.

[0095] It should be noted that the first preset condition is used to determine whether overall performance of the first community reaches an upper limit.

[0096] A specific implementation of S130 includes: the management apparatus determines that a first target feature of the target component meets the first preset condition, where the first target feature is at least one feature specified by the user among a plurality of features of the target component.

[0097] In an embodiment, the first target feature includes at least one of processor (central processing unit, CPU) usage, memory usage, and an input / output (input / output, I / O) throughput.

[0098] When the first target feature includesthe CPU usage, the first preset condition includes: the CPU usage is higher than preset CPU usage (that is, preset first CPU usage). For example, the first preset condition includes that the CPU usage is higher than 90%.

[0099] When the first target feature includes the memory usage, the first preset condition includes that the memory usage is higher than preset memory usage (that is, preset first memory usage). For example, the first preset condition includes that the memory usage is higher than 80%.

[0100] When the first target feature includes the I / O throughput, the first preset condition includes that the I / O throughput is higher than a preset I / O throughput (that is, a preset first I / O throughput). For example, the first preset condition includes that the memory usage is higher than 500 Mbit / s.

[0101] For example, it is assumed that the target component is the cloud service component A, and that the first target feature includes the CPU usage, the memory usage, and the I / Othroughput. In this case, the first preset condition includes: the CPU usage is higher than the preset first CPU usage, the memory usage is higher than the preset first memory usage, and the I / O throughput is higher than the preset first I / O throughput. In this case, it is assumed that the preset first CPU usage is 90%, and the preset first memory usage is 80%, and the preset first I / O throughput is 500 Mbit / s; and current CPU usage of the cloud service component A is 80%, and currentmemory usage of the cloud service component A is 91%, and a current I / O throughput of the cloud service component A is 600 Mbit / s. In this case, because the current CPU usage 80% of the cloud service component A is less than the preset first CPU usage 90%, the target component in the first community does not meet the first preset condition.

[0102] When the first community does not meet the first preset condition, it indicates that current load pressure of the first community is relatively low, so thatoverall performance of the first community is higher than a performance lower limit of the community. Therefore, the management apparatus ends the current method.

[0103] When the target component in the first community meets the first preset condition, the management apparatus performs the following S140.

[0104] S140: The management apparatus creates a second community.

[0105] A call relationship between cloud service components included in the second community is the same as the call relationship between the cloud service components included in the first community. In other words, the second community is the same as the first community, that is, the second community and the first community are the same community.

[0106] An implementation of S140 is to obtain the second community by cloning the first community. A specific implementation of S140 includes: the management apparatus obtains an orchestration template of the first community, 11 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55where the orchestration template includes information such as a component resource type, a relationship between component resources, and the foregoing call relationship, the component resource type includes a resource type (such as a storage resource or a computing resource) required by each of the plurality of cloud service components in the first community, and the relationship between component resources indicates a relationship (for example, a dependency relationship) between resources required by each of the plurality of cloud service components in the first community. Then the management apparatus deploys, by using the orchestration template as a standard, aclone community consistent with the first community, that is, the second community.

[0107] It should be noted that the orchestration template may be obtained by the management apparatus after the management apparatus performs S120, obtains information such as the component resource type of the first community, the relationshipbetween component resources, and the call relationship, and stores the information locally as the orchestration template. Alternatively, the orchestration template may be directly obtained by the management apparatus from the first community when the management apparatus performs S140. A format of the orchestration template is generally a JS object notation (java script object notation, JSON) format or an extensible markup language (extensible markup language, XML) format. Specifically, a manner of obtaining the orchestration template is not specifically limited in this embodiment of this application.

[0108] It should be noted that the second community is configured to share a part of load of the first community. The second community may be specifically configured to share a part of load of some cloud service components in the first community, or the second community may be specifically configured to share a part of load of all the cloud service components in the first community. Thisis specifically determined by the call relationship in the first community and a load balancing policy.

[0109] It should be understood that the second community is completely consistent with the first community. Based on the load balancing policy, the second community is configured to share a part of load of the first community. For example, it is assumed that the load balancing policy is a polling policy, the second community shares half of the load of the first community.

[0110] For example, it is assumed that the first community includes the cloud service component A to the cloud service component C, where the cloud service component A calls the cloud service component B, and the cloud service component B calls the cloud service component C, that is, the cloud service component A→the cloud service component B→the cloud service component C. In this case, when the cloud service component D outside the first community (for example, in a third community) calls the cloud servicecomponent B, because the cloud service component B calls the cloud service component C but does not call the cloud service component A, in this scenario, when the cloud service component D sends a plurality of groups of data to the cloud service component B, a cloud service component B’ and a cloud service component C’ in the second community share a part of load of the cloud service component B and the cloud service component C based on the load balancing policy.

[0111] For another example, based on the foregoing example, the first community includes the cloud service component A→the cloud service component B→the cloud service component C. In this case, when the cloud service component D outside the first community (for example, in the third community) calls the cloud service component A, because the cloud service component A calls the cloud service component B, and the cloud service component B calls the cloud service component C, in this scenario, when the cloud service component Dsends a plurality of groups of data to the cloud service component A, a cloud service component A’, the cloud service component B’, and the cloud service component C’ in the second community share a part of load of the cloud service component A, the cloud service component B, and the cloud service component C in the first community based on the load balancing policy.

[0112] In an embodiment, when overall performance of the first community and the second community is lower than a condition preset by the user, the management apparatus releases the first community or the second community, thereby saving available resources in the cloud platform. For a specific method for releasing the first community or the second community, refer to the conventional technology. Details are not described herein again.

[0113] Based on this, an embodiment of this application provides a cloud platform management method. The method includes: grouping, based on call relationships between a plurality of cloudservice components in a cloud platform, the plurality of cloud service components into a plurality of communities including a first community, where a degree of closeness of a call relationship between cloud service components in a community is relatively high, and a degree of closeness of a call relationship between communities is relatively low; and when the first community meets a preset condition, creating a second community same as the first community (for example, cloning the first community to obtain the second community), so that the second community shares a part of load of the first community. Therefore, when load pressure of the cloud platform is relatively high, an entire community including a plurality of cloud service components is cloned. Because a degree of closeness of a call relationship between cloud service components in the entire community is relatively high, load balancing can be performed on the first community and the second community to quickly adjust the loadpressure of the cloud platform, thereby improving working performance of the cloud platform.

[0114] It should be noted that a call relationship exists between communities. When a degree of closeness of a call relationship between one community (for example, the first community) and another community is relatively high, capacity 12 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 expansion performed only on the first community may cause excessively heavy load pressure on the another community, and consequently, effect of mitigating the load pressure on the entire cloud platform is relatively poor.

[0115] Based on this, in an embodiment, with reference to FIG. 2, as shown in FIG. 5, a specific implementation of S140 further includes S141 to S144.

[0116] S141: The management apparatus obtains modularity between the first community and each of other commu- nities than the first community among the plurality of communities.

[0117] Modularity is a value used to describe a degree of closenessof a call relationship between a cloud service component in one community and a cloud service component in another community. In other words, modularity is a value used to describe a degree of closeness of a call relationship between communities.

[0118] It should be noted that, a specific implementation of S141 may be that after performing S120, the management apparatus stores modularity between any one of the plurality of communities and another community, and when performing S141, the management apparatus directly obtains modularity between the first community and another community locally. Alternatively, a specific implementation of S141 may be that the management apparatus calculates modularity between the first community and each of the other communities based on the foregoing formula (1). Specifically, a specific implementation of S141 is not limited in this embodiment of this application.

[0119] It should be noted that, when a specific implementation of S141 is that themanagement apparatus calculates the modularity between the first community and each of the other communities based on the foregoing formula (1), for a specific calculation process, refer to related descriptions of S120. Details are not described herein again.

[0120] S142: The management apparatus determines, based on the modularity between the first community and each of the other communities, whether a target community exists among the other communities.

[0121] The target community is a community whose modularity with the first community is greater than a threshold among the other communities, that is, a degree of closeness of a call relationship between the target community and the first community is greater than the threshold. In other words, the target community is a community whose degree of closeness of a call relationship with the first community is relatively high among the other communities.

[0122] For example, it is assumed that the plurality of communities include fourcommunities: the first community to a fourth community, where modularity between the first community and the second community is 0.6, modularity between the first community and the third community is 0.2, modularity between the first community and the fourth community is 0.3, and the threshold is 0.55. In this case, because the modularity 0.6 between the first community and the second community is greater than the threshold 0.55, it is determined that the second community is the target community.

[0123] When the target community does not exist among the other communities, the following S143 is performed.

[0124] When the target community exists among the other communities, the following S144 is performed.

[0125] S143: The management apparatus creates the second community.

[0126] It should be noted that an implementation of S143 is consistent with an implementation of S140. For detailed descriptions of S143, refer to the related descriptions of S140. Details are not described hereinagain.

[0127] S144: The management apparatus creates a second community combination.

[0128] A specific implementation of S144 includes: obtaining the second community combination by cloning a first community combination including the first community and the target community.

[0129] The second community combination is the same as the first community combination including the first community and the target community. In other words, the second community combination is a clone combination of the first community combination.

[0130] The second community combination includes the second community and the third community, where the second community is a clone community of the first community, and the third community is a clone community of the target community. In other words, a specific implementation of S144 is to clone the first community and the target community as an entire community to obtain the second community combination.

[0131] It should be understood that a call relationshipbetween the second community and the third community in the second community combination is consistent with a call relationship between the first community and the target community in the first community combination.

[0132] It should be noted that the second community combination is used to share a part of load of the first community combination. The second community combination may be specifically used to share a part of load of some cloud service components in the first community combination, or may be used to share a part of load of all cloud service components in the first community combination. A specific sharing manner is similar to the sharing manner in S140. For details, refer to related descriptions of S140. Details are not described herein again.

[0133] It should be noted that an implementation of S144 is similar to an implementation of S140. For detailed descriptions of S144, refer to the related descriptions of S140. Details are not described herein again.

[0134] In thisembodiment of this application, it is determined whether the target community whose modularity with the first community is greater than the threshold exists among the plurality of communities; and when the target community exists (that is, a community whose degree of closeness of a call relationship with the first community is relatively high exists), the first community and the target community are used as an entire community (that is, the first community 13 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 combination), and the second community combination that includes the second community and the third community and that is the same as the first community combination is created, where the second community combination is used to share load of the first community combination, so as to resolve a problem that load pressure of the target community is excessively high because capacity expansion is performed only on the first community, and further improve effectiveness of reducing the loadpressure of the cloud platform.

[0135] It should be noted that, in S130 and S140, when the target component in the first community meets the first preset condition, capacity expansion is performed at a community level (that is, capacity expansion is performed by cloning a community). When the target component in the first community does not meet the first preset condition, load of at least one cloud service component and link in the first community may be relatively heavy, and in severe cases, the entire cloud platform may break down.

[0136] Based on this, in an embodiment, when the management apparatus determines that the target component in the first community does not meet the first preset condition in S130, as shown in FIG. 6, the method further includes S210 and S220.

[0137] S210: The management apparatus determines a hot point and a hot link based on a call relationship between cloud service components in the first community.

[0138] The hot point is a cloud service component witha relatively high call frequency (that is, a popularity value) among the plurality of cloud service components in the first community, where a popularity value of a cloud service component is a sum of an out-degree of the cloud service component and an in-degree of the cloud service component. The hot link a link with a relatively high call frequency between the plurality of cloud service components in the first community.

[0139] In an example, a manner of determining the hot point and the hot link is as follows:

[0140] A formula for determining the hot point in the first community is shown in the following formula (2):

[0141] In the formula, Node is a hot point; max represents a maximum value; rij represents a quantity of times that a node i (that is, a cloud service component i) calls a node j; and x is a total quantity of nodes (that is, cloud service components) in the first community.

[0142] The hot link is a link with a relatively high call frequency between the plurality ofcloud service components in the first community. A formula for determining the hot link is shown in the following formula (3):

[0143] In the formula, Link is a hot link; max represents a maximum value; rij represents a quantity of times that the node i calls the node j, that is, a call frequency of the link used when the node i calls the node j; and x is the total quantity of nodes in the first community.

[0144] For example, in matrix data corresponding to the first community shown in FIG. 7, rij(x) indicates that the node i calls the node j for x times. The first community includes five nodes, and each node represents one cloud service component. Call data of the five nodes specifically includes {r11(2), r12(1), r13(2), r14(2), r15(2)}, {r21(1), r22(1), r23(5), r24(3), r25(8)}, {r31(3), r32(4), r33(0), r34(2), r35(6)}, {r41(2), r42(1), r43(3), r44(0), r45(2)}, and {r51(7), r52(2), r53(2), r54(9), r55(5)}. Popularity values of a node 1 to a node 5 are calculated separately based on theforegoing formula (2), and the obtained popularity values of the node 1 to the node 5 are 22, 26, 27, 24, and 43 in sequence. In other words, it is determined that the node 5 with a highest popularity value is a hot point, and it is determined that a link corresponding to r54 with a highest call frequency is a hot link.

[0145] In another example, a specific implementation of S210 includes S1 to S3.

[0146] S1: The management apparatus obtains respective popularity values of the plurality of cloud service compo- nents in the first community and call frequencies of a plurality of links between the plurality of cloud service components.

[0147] It should be noted that the management apparatus obtains the respective popularity values of the plurality of cloud service components through calculation based on the foregoing formula (2); and the management apparatus determines the call frequencies of the plurality of links based on the call relationships of the cloud service components in thefirst community.

[0148] S2: The management apparatus determines that top N cloud service components with highest popularity values are hot points.

[0149] N is a preset integer greater than 0.

[0150] For example, based on the example of S210, it is assumed that when N is 2, because the popularity values of the node 1 to the node 5 are 22, 26, 27, 24, and 43 in sequence, the management apparatus determines that the node 5 and the node 3 are hot points.

[0151] S3: The management apparatus determines that top M links with highest call frequencies are hot links. 14 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55

[0152] M is a preset integer greater than 0.

[0153] For example, based on the example of S210, assuming that M is 2, in this case, the management apparatus determines that links (r54 for short) corresponding to r54 and r15 are hot links.

[0154] It should be noted that, after determining the hot point and the hot link in the first community, the management apparatus may mark the hotpoint and the hot link in the call network corresponding to the first community, so that when viewing the call network, the user can intuitively learn a distribution status of the hot point and the hot link in the first community, thereby improving user experience.

[0155] S220: The management apparatus processes the hot point and the hot link based on a preset mitigation policy.

[0156] The preset mitigation policy is used to enable the processed hot point and hot link to meet respective current load requirements. In other words, the preset mitigation policy is a method for reducing load pressure of the hot point and the hot link.

[0157] S220 includes two specific implementations, which are specifically as follows:

[0158] In an implementation, as shown in FIG. 8, a specific implementation of S220 includes S220a to S220d.

[0159] S220a: The management apparatus determines a target hot point mitigation policy from a first target corre- spondence based on an identifier of a community inwhich the hot point is located and a second target feature of the hot point.

[0160] The second target feature is a feature specified in advance by the user among a plurality of features of the hot point, and the second target feature includes at least one of CPU usage, memory usage, and an I / O throughput.

[0161] It should be understood that the second target feature is a feature specified in advance by the user among the plurality of features of the hot point based on a specific function (that is, a service type provided for the cloud platform) of the hot point. For example, when the hot point is a cloud service component used to store data, the second target feature may include storage space usage. For another example, when the hot point is a cloud service component used for data calculation, the second target feature may include CPU usage.

[0162] It should be noted that in this embodiment of this application, an example in which the second target feature includes the CPU usage, thememory usage, and the I / O throughput is used for description. Details are not described subsequently.

[0163] The first target correspondence includes a correspondence between identifiers of the plurality of communities, a plurality of third preset conditions, and a plurality of hot point mitigation policies. The first target correspondence is shown in Table 1, and the identifiers of the plurality of communities include the first community and the second community. The plurality of third preset conditions include: "CPU usage > preset second CPU usage, memory usage > preset second memory usage, and I / O throughput > preset second I / O throughput"; "CPU usage > preset third CPU usage, memory usage > preset third memory usage, and I / O throughput > preset third I / O throughput"; and "CPU usage > preset fourth CPU usage, and I / O throughput > preset fourth I / O throughput". The plurality of hot point mitigation policies include vertical scaling and horizontal scaling. The third preset conditionis a condition that the second target feature needs to meet, and the third preset condition is used to determine whether to expand a capacity of the hot point. Table 1 Community identifier Third preset condition Hot point mitigation policy First community CPU usage > preset second CPU usage, memory usage > preset second memory usage, and ImplementationI / O throughput > preset second I / O throughput Horizontal scaling First community CPU usage > preset third CPU usage, memory usage > preset third memory usage, and ImplementationI / O throughput > pre- set third I / O throughput Vertical scaling Second community CPU usage > preset fourth CPU usage, and ImplementationI / O throughput > preset fourth I / O throughput Vertical scaling

[0164] The vertical scaling is to provide concurrency by improving hardware performance of a single machine, and is specifically to increase a quantity of CPUs in the hot point, and / or increase a capacity of memory in the hot point (for example, add a memory module),and / or add a data bus of the hot point.

[0165] The horizontal scaling is to increase a quantity of cloud service components represented by the hot point, and reduce load pressure of the hot point by using a load balancing policy. For example, when the hot point is a cloud service component A, a quantity of cloud service components A is increased.

[0166] It should be noted that, for specific implementations of the vertical scaling and the horizontal scaling, refer to the conventional technology. Details are not described herein again.

[0167] The identifier of the community corresponding to the target hot point mitigation policy is consistent with the 15 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 identifier of the first community in which the hot point is located, and the second target feature of the hot point meets the third preset condition corresponding to the target hot point mitigation policy.

[0168] For example, assuming that the identifier of the community in which the hotpoint is located is the first community, CPU usage of the hot point > preset third CPU usage, memory usage > preset third memory usage, and I / O throughput > preset third I / O throughput. In this case, the target hot point mitigation policy of the hot point is vertical scaling.

[0169] S220b: The management apparatus processes the hot point based on the target hot point mitigation policy.

[0170] For example, based on the example in S220a, performing vertical scaling on the hot point specifically includes: increasing a quantity of CPUs of the hot point, increasing a memory storage capacity of the hot point, and increasing a quantity of data buses of the hot point.

[0171] S220c: The management apparatus determines a target hot link mitigation policy from a second target correspondence based on a third target feature of the hot link.

[0172] The third target feature is a feature specified in advance by the user among a plurality of features of the hot link. For example, the third targetfeature includes a data transmission success rate and an I / O throughput.

[0173] It should be noted that in this embodiment of this application, an example in which the third target feature includes the I / O throughput is used for description. Details are not described subsequently.

[0174] A fourth preset condition is acondition that the third target featureneeds to meet, and the fourth preset condition is used to determine whether to perform scaling on the hot link.

[0175] The second target correspondence includes a correspondence between a plurality of fourth preset conditions and a plurality of hot link mitigation policies. Specifically, as shown in Table 2, the plurality of fourth preset conditions include: "ImplementationI / O throughput > 600 Mbit / s, and I / O throughput≥400 Mbit / s" and "ImplementationI / O throughput > 400 Mbit / s". The plurality of hot link mitigation policies include lateral scaling and longitudinal scaling. Table 2 Fourth preset condition Hot link mitigation policyImplementationI / O throughput > 600 Mbit / s, and I / O throughput ≥ 400 Mbit / s Lateral scaling ImplementationI / O throughput > 400 Mbit / s Longitudinal scaling

[0176] The lateral scaling may be specifically increasing a quantity of links, so that a part of load of the hot link is shared by added links.

[0177] The longitudinal scaling may be specifically increasing a bandwidth of the hot link to increase an amount of data that can be transmitted by the hot link.

[0178] A specific implementation of S220c includes: the management apparatus determines a hot link mitigation policy corresponding to a fourth preset condition that the third target feature of the hot link meets as the target hot link mitigation policy.

[0179] For example, assuming that a current I / O throughput of the hot link is 550 Mbit / s, in this case, it is determined that the hot link mitigation policy (that is, lateral scaling) corresponding to the fourth preset condition including Implementa- tionI / O throughput > 600 Mbit / s andI / O throughput ≥ 400 Mbit / s is the target hot link mitigation policy.

[0180] S220d: The management apparatus processes the hot link based on the target hot link mitigation policy.

[0181] For example, based on the example in S220c, the target hot link mitigation policy is lateral scaling, and a quantity of hot links is increased, so that added links share a part of load of the hot link.

[0182] In the foregoing embodiment, when the target component in the first community does not meet the first preset condition, the hot point and the hot link in the first community are determined; and then the target hot point mitigation policy is determined based on the community in which the hot point is located and the third preset condition that the second target feature of the hot point meets, and the hot point is processed based on the target hot point mitigation policy, so that load pressure of the hot point is mitigated. Similarly, the management apparatus determines that the hot link mitigationpolicy corresponding to the fourth preset condition that the third target feature of the hot link meets is the target hot link mitigation policy, and processes the hot link based on the target hot link mitigation policy, to reduce load pressure of the processed hot link, thereby resolving a problem that the cloud platform breaks down due to relatively heavy load of at least one cloud service component and link in the first community.

[0183] In another implementation, as shown in FIG. 9, a specific implementation of S220 includes S220A to S220C.

[0184] S220A: The management apparatus determines a target type template from a plurality of preset type templates based on a distribution feature of the hot point and the hot link.

[0185] The distribution feature of the hot point and the hot link is used to represent spatial location information of the hot point and the hot link in the first community. In other words, the distribution feature indicates location information of the hot point andthe hot link in the first community.

[0186] Respective features of nodes and links in any two types of templates among the plurality of preset type templates 16 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 are different, and the target type template is a type template having a highest degree of matching with the distribution feature of the hot point and the hot link among the plurality of preset type templates.

[0187] A specific implementation of S220A includes: the management apparatus separately calculates, based on the distribution feature of the hot point and the hot link, a degree of matching with each preset type template; and then the management apparatus determines that a type template with a highest degree of matching is the target type template.

[0188] It should be noted that the management apparatus may calculate, based on a training model, a degree of matching between the distribution feature of the hot point and the hot link and each preset type template. Alternatively,the management apparatus may calculate, based on another existing algorithm, a degree of matching between the distribution feature of the hot point and the hot link and each preset type template. Specifically, a specific implementation of S220A is not limited in this embodiment of this application.

[0189] S220B: The management apparatus determines, from a correspondence between a plurality of type templates and a plurality of mitigation policies, a target mitigation policy corresponding to the target type template.

[0190] The plurality of type templates include a plurality of different types of templates, for example, a centralized template and a distributed template.

[0191] The correspondence is shown in Table3. The plurality of type templates include a type template 1,a type template 2, and a type template 3. The plurality of mitigation policies include: expanding the capacity of the hot point, expanding the capacity of the hot point and the capacity of the hot link, and expandingthe capacity of the hot link. Table 3 Type template Mitigation policy Type template 1 Expanding the capacity of the hot point Type template 2 Expanding the capacity of the hot point and the capacity of the hot link Type template 3 Expanding the capacity of the hot link

[0192] It should be noted that a specific manner of expanding the capacity of the hot point may be vertical scaling or horizontal scaling, and that a specific manner of expanding the capacity of the hot link may be longitudinal scaling or lateral scaling. Specifically, the specific manner of expanding the capacity of the hot point and the specific manner of expanding the capacity of the hot link are not limited in this embodiment of this application.

[0193] For example, the type template 1 in Table 3 is a centralized template in FIG. 10, the type template 2 is a link template in FIG. 10, and the type template 3 is a distributed template in FIG. 10. When the target type template is the centralized template, it isdetermined that the target mitigation policy is expanding the capacity of the hot point.

[0194] S220C: The management apparatus processes the hot point and the hot link based on the target mitigation policy.

[0195] In the foregoing embodiment, when the target component in the first community does not meet the first preset condition, the hot point and the hot link in the first community are determined; and then the management apparatus determines the target type template from the plurality of preset type templates based on the distribution feature of the hot point and the hot link, and processes the hot point and the hot link based on the target mitigation policy corresponding to the target type template, to reduce load pressure of the processed hot point and hot link, thereby resolving a problem that the cloud platform breaks down due to relatively heavy load of at least one cloud service component and link in the first community.

[0196] The foregoing mainly describes the solutionsprovided in embodiments of this application from the perspective of the method. To implement the foregoing functions, the management apparatus includes corresponding hardware structures and / or software modules for performing the functions. Persons skilled in the art should easily be aware that, in combination with units and algorithm steps of the examples described in embodiments disclosed in this specification, this application may be implemented by hardware or a combination of hardware and computer software. Whether a function is performed by hardware or hardware driven by computer software depends on particular applications and design constraints of the technical solutions. Persons skilled in the art may use different methods to implement the described functions for each particular application, but it should not be considered that the implementation goes beyond the scope of this application.

[0197] In embodiments of this application, the management apparatus may be divided intofunctional modules based on the foregoing method examples. For example, the management apparatus may include each functional module obtained through division based on each corresponding function, or two or more functions may be integrated into one processing module. The integrated module may be implemented in a form of hardware, or may be implemented in a form of a software functional module. It should be noted that the module division in embodiments of this application is an example, and is merely logical function division. There may be another division manner during actual implementation.

[0198] When functional modules are obtained through division based on corresponding functions, FIG. 11 is a diagram 17 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 of a possible structure of the management apparatus in the foregoing embodiment. As shown in FIG. 11, the management apparatus includes a grouping module 1101 and a creation module 1102.

[0199] The grouping module 1101 is configured togroup a plurality of cloud service components in a cloud platform into a plurality of communities based on call relationships between the plurality of cloud service components, for example, perform step S120 in the foregoing method embodiment.

[0200] The creation module 1102 is configured to create a second community if a target component in a first community of the plurality of communities meets a first preset condition, for example, perform step S140 in the foregoing method embodiment.

[0201] Optionally, the grouping module 1101 is configured to group the plurality of cloud service components into the plurality of communities based on a community discovery algorithm and the call relationships.

[0202] Optionally, the creation module 1102 is configured to create the second community if a first target feature of the target component meets the first preset condition.

[0203] Optionally, the management apparatus further includes an obtaining module 1103 and a determining module 1104.

[0204] The obtaining module 1103 is configured to obtain modularity between the first community and each of other communities than the first community among the plurality of communities, for example, perform step S141 in the foregoing method embodiment.

[0205] The determining module 1104 is configured to determine, based on the modularity, whether a target community exists among the other communities, for example, perform step S142 in the foregoing method embodiment.

[0206] The creation module 1102 is configured to create the second community when the target community does not exist among the other communities, for example, perform step S 143 in the foregoing method embodiment.

[0207] Optionally, the creation module 1102 is further configured to create a second community combination when the target community exists among the other communities, for example, perform step S144 in the foregoing method embodiment.

[0208] Optionally, the determining module 1104 is configured to determine ahot point and a hot link based on a call relationship between cloud service components in the first community if the target component does not meet the first preset condition, for example, perform step S210 in the foregoing method embodiment; and the determining module 1104 is further configured to process the hot point and the hot link based on a preset mitigation policy, for example, perform step S220 in the foregoing method embodiment.

[0209] Optionally, the determining module 1104 is configured to determine a target hot point mitigation policy from a first target correspondence based on an identifier of a community in which the hot point is located and a second target feature of the hot point, for example, perform step S220a in the foregoing method embodiment; and the determining module 1104 is configured to process the hot point based on the target hot point mitigation policy, for example, perform step S220b in the foregoing method embodiment.

[0210] Optionally, the determiningmodule 1104 is configured to determine a target hot link mitigation policy from a second target correspondence based on a third target feature of the hot link, for example, perform step S220c in the foregoing method embodiment; and the determining module 1104 is further configured to process the hot link based on the target hot link mitigation policy, for example, perform step S220d in the foregoing method embodiment.

[0211] Optionally, the determining module 1104 is configured to determine a target type template from a plurality of preset type templates based on a distribution feature of the hot point and the hot link, for example, perform step S220A in the foregoing method embodiment; the determining module 1104 is configured to determine, from a correspondence between a plurality of type templates and a plurality of mitigation policies, a target mitigation policy corresponding to the target type template, for example, perform step S220B in the foregoing method embodiment; and thedetermining module 1104 is further configured to process the hot point and the hot link based on the target mitigation policy, for example, perform step S220C in the foregoing method embodiment.

[0212] The grouping module 1101, the creation module 1102, the obtaining module 1103, and the determining module 1104 may all be implemented by using software, or may be implemented by using hardware. For example, the following uses the grouping module 1101 as an example to describe an implementation of the grouping module 1101. Similarly, for implementations of the creation module 1102, the obtaining module 1103, and the determining module 1104, refer to the implementation of the grouping module 1101.

[0213] A module is used as an example of a software functional unit, and the grouping module 1101 may include code run on a computing instance. The computing instance may include at least one of a physical host (computing device), a virtual machine, and a container. Further, there may be one ormore computing instances. For example, the grouping module 1101 may include code run on a plurality of hosts, virtual machines, or containers. It should be noted that the 18 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 plurality of hosts, virtual machines, or containers configured to run the code may be distributed in a same region (region), or may be distributed in different regions. Further, the plurality of hosts, virtual machines, or containers configured to run the code may be distributed in a same availability zone (availability zone, AZ), or may be distributed in different AZs. Each AZ includes one data center or a plurality of data centers that are geographically close to each other. Usually, one region may include a plurality of AZs.

[0214] Similarly, the plurality of hosts, virtual machines, or containers configured to run the code may be distributed in a same virtual private cloud (virtual private cloud, VPC), or may be distributed in a plurality of VPCs. Usually, one VPCis arranged in one region. For cross-region communication between two VPCs in a same region and between VPCs in different regions, a communication gateway needs to be arranged in each of the VPCs, and interconnection between the VPCs is implemented through the communication gateway.

[0215] A module is used as an example of a hardware functional unit, and the grouping module 1101 may include at least one computing device, for example, a server. Alternatively, the grouping module 1101 may be a device implemented by using an application-specific integrated circuit (application-specific integrated circuit, ASIC) or a programmable logic device (programmable logic device, PLD), or the like. The PLD may be implemented by a complex programmable logic device (complex programmable logic device, CPLD), a field-programmable gate array (field-programmable gate array, FPGA), generic array logic (generic array logic, GAL), or any combination thereof.

[0216] A plurality of computing devices includedin the grouping module 1101 may be distributed in a same region, or may be distributed in different regions. The plurality of computing devices included in the grouping module 1101 may be distributed in a same AZ, or may be distributed in different AZs. Similarly, the plurality of computing devices included in the grouping module 1101 may be distributed in a same VPC, or may be distributed in a plurality of VPCs. The plurality of computing devices may be any combination of computing devices such as a server, an ASIC, a PLD, a CPLD, an FPGA, and a GAL.

[0217] It should be noted that in another embodiment, the grouping module 1101 may be configured to perform any step in the foregoing cloud platform management method; the creation module 1102 may be configured to perform any step in the foregoing cloud platform management method; the obtaining module 1103 may be configured to perform any step in the foregoing cloud platform management method; and the determining module 1104 may beconfigured to perform any step in the foregoing cloud platform management method. The steps implemented by the grouping module 1101, the creation module 1102, the obtaining module 1103, and the determining module 1104 may be specified as required. The grouping module 1101, the creation module 1102, the obtaining module 1103, and the determining module 1104 respectively implement different steps in the foregoing cloud platform management method to implement all functions of the management apparatus.

[0218] An embodiment of this application further provides a computing device cluster. The computing device cluster includes at least one computing device. The computing device may be a server, for example, a central server, an edge server, or a local server in a local data center. In some embodiments, the computing device may alternatively be a terminal device, for example, a desktop computer, a notebook computer, or a smartphone.

[0219] As shown in FIG. 12, the computing device clusterincludes at least one computing device 100. A memory or memories 106 in one or more computing devices 100 in the computing device cluster may store same instructions for performing the foregoing cloud platform management method.

[0220] In some possible implementations, alternatively, a memory or memories 106 in one or more computing devices 100 in the computing device cluster each may store a part of instructions for performing the foregoing cloud platform management method. In other words, a combination of the one or more computing devices 100 may jointly execute instructions used to perform the foregoing cloud platform management method.

[0221] It should be noted that memories 106 in different computing devices 100 in the computing device cluster may store different instructions respectively used to perform some functions of the management apparatus. In other words, instructions stored in the memories 106 in different computing devices 100 may implement functions of one or more ofthe grouping module 1101, the creation module 1102, the obtaining module 1103, and the determining module 1104.

[0222] In some possible implementations, the one or more computing devices in the computing device cluster may be connected through a network. The network may be a wide area network, a local area network, or the like. FIG. 13 shows a possible implementation. As shown in FIG. 13, two computing devices 100A and 100B are connected through a network. Specifically, each computing device is connected to the network through a communication interface of the computing device. In this possible implementation, a memory 106 in the computing device 100A stores instructions for executing functions of the grouping module 1101 and the creation module 1102. In addition, a memory 106 in the computing device 100B stores instructions for performing functions of the obtaining module 1103 and the determining module 1104.

[0223] A requirement for obtaining modularity between a first community andeach of other communities than the first community among a plurality of communities and determining a target community among the other communities in a cloud platform management method provided in this application may be considered for a connection manner between computing device clusters shown in FIG. 13. Therefore, it is considered that functions implemented by the obtaining module 1103 and the determining module 1104 are performed by the computing device 100B. 19 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55

[0224] It should be understood that functions of the computing device 100A shown in FIG. 13 may also be completed by a plurality of computing devices 100. Similarly, functions of the computing device 100B may also be completed by a plurality of computing devices 100.

[0225] An embodiment of this application further provides another computing device cluster. For a connection relation- ship between computing devices in the computing device cluster, refer to the connection mannerin the computing device cluster in FIG. 4 and FIG. 5 similarly. A difference lies in that a memory or memories 106 in one or more computing devices 100 in the computing device cluster may store same instructions for performing the cloud platform management method.

[0226] In some possible implementations, alternatively, a memory or memories 106 in one or more computing devices 100 in the computing device cluster each may store a part of instructions for performing the cloud platform management method. In other words, a combination of the one or more computing devices 100 may jointly execute instructions used to perform the cloud platform management method.

[0227] An embodiment of this application further provides a computer program product including instructions. The computer program product may be a software or program product that includes instructions and that can be run on a computing device or be stored in any usable medium. When the computer program product is run on at least onecomputing device, the at least one computing device is enabled to perform the cloud platform management method.

[0228] An embodiment of this application further provides a computer-readable storage medium. The computer- readable storage medium may be any usable medium that can be stored by a computing device, or a data storage device, such as a data center, including one or more usable media. The usable medium may be a magnetic medium (for example, a floppy disk, a hard disk, or a magnetic tape), an optical medium (for example, a DVD), a semiconductor medium (for example, a solid-state drive), or the like. The computer-readable storage medium includes instructions. The instructions instruct a computing device to perform a cloud platform management method.

[0229] Finally, it should be noted that the foregoing embodiments are merely intended for describing the technical solutions of the present invention, but not for limiting the present invention. Although the present invention isdescribed in detail with reference to the foregoing embodiments, persons of ordinary skill in the art should understand that they may still make modifications to the technical solutions described in the foregoing embodiments or make equivalent replacements to some technical features thereof, without departing from the scope of the technical solutions of embodiments of the present invention. Claims 1. A cloud platform management method, comprising: grouping a plurality of cloud service components in a cloud platform into a plurality of communities based on call relationships between the plurality of cloud service components, wherein each community comprises at least two of the plurality of cloud service components, a degree of closeness of a call relationship between cloud service components comprised in each community is greater than or equal to a preset degree of closeness, and a degree of closeness between a cloud service component in any one of the plurality of communities and acloud service component in another one of the plurality of communities is less than the preset degree of closeness; and if a target component in a first community of the plurality of communities meets a first preset condition, creating a second community, wherein the second community is the same as the first community, the second community is configured to share a part of load of the first community, the first preset condition indicates that performance of the target component is lower than preset performance, the first community is any one of the plurality of communities, and the target component is at least one of a plurality of cloud service components in the first community. 2. The method according to claim 1, wherein the call relationship comprises one or more of the following: call information of the plurality of cloud service components in the cloud platform, a call frequency of a link between any two of the plurality of cloud service components in the cloud platform, and anout-degree and an in-degree of any one of the plurality of cloud service components in the cloud platform. 3. The method according to claim 1 or 2, wherein grouping the plurality of cloud service components in the cloud platform into the plurality of communities based on the call relationships between the plurality of cloud service components comprises: grouping the plurality of cloud service components into the plurality of communities based on a community discovery algorithm and the call relationships. 4. The method according to any one of claims 1 to 3, wherein the target component comprises: a cloud service component that meets a second preset condition among the plurality 20 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 of cloud service components in the first community, or at least one cloud service component specified by a user among the plurality of cloud service components in the first community. 5. The method according to claim 4,wherein thesecond preset condition comprisestop X cloud service components with highest popularity values among the plurality of cloud service components in the first community, wherein the popularity value is a sum of an out-degree of a cloud service component and an in-degree of the cloud service component, and X is an integer greater than 0. 6. The method according to any one of claims 1 to 5, wherein creating the second community if the target component in the first community of the plurality of communities meets the first preset condition comprises: if a first target feature of the target component meets the first preset condition, creating the second community, wherein the first target feature is at least one feature specified by the user among a plurality of features of the target component. 7. The method according to claim 6, wherein the first target feature comprises at least one of CPU usage, memory usage, and an I / O throughput. 8. The method according to claim 7, wherein when the first target feature comprises the CPUusage, the first preset condition comprises: the CPU usage is higher than preset CPU usage; when the first target feature comprises the memory usage, the first preset condition comprises: the memory usage is higher than preset memory usage; or when the first target feature comprises the I / O throughput, the first preset condition comprises: the I / O throughput is higher than a preset I / O throughput. 9. The method according to any one of claims 1 to 8, wherein when creating the second community, the method further comprises: obtaining modularity between the first community and each of other communities than the first community among the plurality of communities, wherein the modularity is a value used to describe a degree of closeness of a call relationship between communities; determining, based on the modularity, whether a target community exists among the other communities, wherein the target community is a community whose modularity with the first community is greater than a thresholdamong the other communities; and when the target community does not exist among the other communities, creating the second community. 10. The method according to claim 9, wherein the method further comprises: when the target community exists among the other communities, creating a second community combination, wherein the second community combination is the same as a first community combination comprising the first community and the target community, the second community combination comprises the second community and a third community, the third community is the same as the target community, and the second community combination is used to share a part of load of the first community combination. 11. The method according to any one of claims 1 to 10, wherein the method further comprises: if the target component does not meet the first preset condition, determining a hot point and a hot link based on a call relationship between cloud service components in the first community, wherein thehot point is a cloud service component with a relatively high call frequency among the plurality of cloud service components in the first community, and the hot link is a link with a relatively high call frequency between the plurality of cloud service components; and processing the hot point and the hot link based on a preset mitigation policy, wherein the preset mitigation policy is used to enable the processed hot point and hot link to meet respective current load requirements. 12. The method according to claim 11, wherein the hot point is top N cloud service components with highest popularity values among the plurality of cloud service components comprised in the first community, wherein the popularity value is a sum of an out-degree of a cloud 21 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 service component and an in-degree of the cloud service component, and N is an integer greater than 0; and the hot link is top M links with highest call frequencies among a plurality oflinks between the plurality of cloud service components comprised in the first community, wherein M is an integer greater than 0. 13. The method according to claim 11 or 12, wherein processing the hot point based on the preset policy comprises: determining a target hot point mitigation policy from a first target correspondence based on an identifier of a community in which the hot point is located and a second target feature of the hot point, wherein the first target correspondence comprises a correspondence between identifiers of the plurality of communities, a plurality of third preset conditions, and a plurality of hot point mitigation policies, the second target feature is a feature specified in advance by the user among a plurality of features of the hot point, and the third preset condition is used to determine whether to perform capacity expansion on the hot point; and processing the hot point based on the target hot point mitigation policy. 14. The method according to claim 13,wherein the hot point mitigation policy comprises: reset scaling and / or horizontal scaling. 15. The method according to any one of claims 11 to 14, wherein processing the hot link based on the preset policy comprises: determining a target hot link mitigation policy from a second target correspondence based on a third target feature of the hot link, wherein the second target correspondence comprises a correspondence between a plurality of fourth preset conditions and a plurality of hot link mitigation policies, and the third target feature is a feature specified in advance by the user among a plurality of features of the hot link; and processing the hot link based on the target hot link mitigation policy. 16. The method according to claim 15, wherein the hot link mitigation policy comprises: lateral scaling and / or longitudinal scaling. 17. The method according to claim 11 or 12, wherein processing the hot point and the hot link based on the preset policy comprises: determining a targettype template from a plurality of preset type templates based on a distribution feature of the hot point and the hot link, wherein the distribution feature is used to represent spatial location information of the hot point and the hot link, and the target type template is a type template having a highest degree of matching with the distribution feature among the plurality of preset type templates; determining, from a correspondence between a plurality of type templates and a plurality of mitigation policies, a target mitigation policy corresponding to the target type template; and processing the hot point and the hot link based on the target mitigation policy. 18. A management apparatus, wherein the management apparatus comprises a grouping module and a creation module, wherein the grouping module is configured to group a plurality of cloud service components in a cloud platform into a plurality of communities based on call relationships between the plurality of cloud servicecomponents, wherein each community comprises at least two of the plurality of cloud service components, a degree of closeness of a call relationship between cloud service components comprised in each community is greater than or equal to a preset degree of closeness, and a degree of closeness between a cloud service component in any one of the plurality of communities and a cloud service component in another one of the plurality of communities is less than the preset degree of closeness; and the creation module is configured to create a second community if a target component in a first community of the plurality of communities meets a first preset condition, wherein the second community is the same as the first community, the second community is configured to share a part of load of the first community, the first preset condition indicates that performance of the target component is lower than preset performance, the first community is any one of the plurality of communities, and thetarget component is at least one of a plurality of cloud service components in the first community. 19. The management apparatus according to claim 18, wherein 22 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 the call relationship comprises one or more of the following: call information of the plurality of cloud service components in the cloud platform, a call frequency of a link between any two of the plurality of cloud service components in the cloud platform, and an out-degree and an in-degree of any one of the plurality of cloud service components in the cloud platform. 20. The management apparatus according to claim 18 or 19, wherein the grouping module is specifically configured to group the plurality of cloud service components into the plurality of communities based on a community discovery algorithm and the call relationships. 21. The management apparatus according to any one of claims 18 to 20, wherein the target component comprises: a cloud service component that meets asecond preset condition among the plurality of cloud service components in the first community, or at least one cloud service component specified by a user among the plurality of cloud service components in the first community. 22. The management apparatus according to claim 21, wherein the second preset condition comprises top X cloud service components with highest popularity values among the plurality of cloud service components in the first community, wherein the popularity value is a sum of an out-degree of a cloud service component and an in-degree of the cloud service component, and X is an integer greater than 0. 23. The management apparatus according to any one of claims 18 to 22, wherein the creation module is configured to create the second community if a first target feature of the target component meets the first preset condition, wherein the first target feature is at least one feature specified by the user among a plurality of features of the target component. 24. Themanagement apparatus according to claim 23, wherein the first target feature comprises at least one of processor CPU usage, memory usage, and an I / O throughput. 25. The management apparatus according to claim 24, wherein when the first target feature comprises the CPU usage, the first preset condition comprises: the CPU usage is higher than preset CPU usage; when the first target feature comprises the memory usage, the first preset condition comprises: the memory usage is higher than preset memory usage; or when the first target feature comprises the I / O throughput, the first preset condition comprises: the I / O throughput is higher than a preset I / O throughput. 26. The management apparatus according to any one of claims 18 to 25, wherein the management apparatus further comprises an obtaining module and a determining module, wherein the obtaining module is configured to obtain modularity between the first community and each of other communities than the first community among theplurality of communities, wherein the modularity is a value used to describe a degree of closeness of a call relationship between communities; the determining module is configured to determine, based on the modularity, whether a target community exists among the other communities, wherein the target community is a community whose modularity with the first community is greater than a threshold among the other communities; and the creation module is configured to create the second community when the target community does not exist among the other communities. 27. The management apparatus according to claim 26, wherein the creation module is configured to create a second community combination when the target community exists among the other communities, wherein the second community combination is the same as a first community combination comprising the first community and the target community, the second community combination comprises the second community and a third community, the thirdcommunity is the same as the target community, and the second community combination is used to share a part of load of the first community combination. 28. The management apparatus according to any one of claims 18 to 27, wherein the determining module is configured to determine a hot point and a hot link based on a call relationship between 23 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 cloud service components in the first community if the target component does not meet the first preset condition, wherein the hot point is a cloud service component with a relatively high call frequency among the plurality of cloud service components in the first community, and the hot link is a link with a relatively high call frequency between the plurality of cloud service components; and the determining module is further configured to process the hot point and the hot link based on a preset mitigation policy, wherein the preset mitigation policy is used to enable the processed hot point and hotlink to meet respective current load requirements. 29. The management apparatus according to claim 28, wherein the hot point is top N cloud service components with highest popularity values among the plurality of cloud service components comprised in the first community, wherein the popularity value is a sum of an out-degree of a cloud service component and an in-degree of the cloud service component, and N is an integer greater than 0; and the hot link is top M links with highest call frequencies among a plurality of links between the plurality of cloud service components comprised in the first community, wherein M is an integer greater than 0. 30. The management apparatus according to claim 27 or 28, wherein the determining module is configured to determine a target hot point mitigation policy from a first target correspondence based on an identifier of a community in which the hot point is located and a second target feature of the hot point, wherein the first target correspondencecomprises a correspondence between identifiers of the plurality of communities, a plurality of third preset conditions, and a plurality of hot point mitigation policies, the second target feature is a feature specified in advance by the user among a plurality of features of the hot point, and the third preset condition is used to determine whether to perform capacity expansion on the hot point; and the determining module is further configured to process the hot point based on the target hot point mitigation policy. 31. The management apparatus according to claim 30, wherein the hot point mitigation policy comprises: reset scaling and / or horizontal scaling. 32. The management apparatus according to any one of claims 28 to 31, wherein the determining module is configured to determine a target hot link mitigation policy from a second target correspondence based on a third target feature of the hot link, wherein the second target correspondence comprises a correspondence between aplurality of fourth preset conditions and a plurality of hot link mitigation policies, and the third target feature is a feature specified in advance by the user among a plurality of features of the hot link; and the determining module is further configured to process the hot link based on the target hot link mitigation policy. 33. The management apparatus according to claim 32, wherein the hot link mitigation policy comprises: lateral scaling and / or longitudinal scaling. 34. The management apparatus according to claim 28 or 29, wherein the determining module is configured to determine a target type template from a plurality of preset type templates based on a distribution feature of the hot point and the hot link, wherein the distribution feature is used to represent spatial location information of the hot point and the hot link, and the target type template is a type template having a highest degree of matching with the distribution feature among the plurality of preset typetemplates; and the determining module is further configured to determine, from a correspondence between a plurality of type templates and a plurality of mitigation policies, a target mitigation policy corresponding to the target type template; and process the hot point and the hot link based on the target mitigation policy. 35. A computing device cluster, comprising at least one computing device, wherein each computing device comprises a processor and a memory; and the processor of the at least one computing device is configured to execute instructions stored in the memory of the at least one computing device, so that the computing device cluster performs the method according to any one of claims 1 to 17. 24 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 36. A computer-readable storage medium, storing computer instructions, wherein when the computer instructions are run on a computing device, the computing device is enabled to perform the method according to any one of claims 1 to 17.37. A computer program product comprising instructions, wherein when the instructions are run by a computing device cluster, the computing device cluster is enabled to perform the method according to any one of claims 1 to 17. 25 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 26 EP 4 697 678 A1 27 EP 4 697 678 A1 28 EP 4 697 678 A1 29 EP 4 697 678 A1 30 EP 4 697 678 A1 31 EP 4 697 678 A1 32 EP 4 697 678 A1 33 EP 4 697 678 A1 34 EP 4 697 678 A1 35 EP 4 697 678 A1 36 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 37 EP 4 697 678 A1 5 10 15 20 25 30 35 40 45 50 55 38 EP 4 697 678 A1 REFERENCES CITED IN THE DESCRIPTION This list of references cited by the applicant is for the reader’s convenience only. It does not form part of the European patent document. Even though great care has been taken in compiling the references, errors or omissions cannot be excluded and the EPO disclaims all liability in this regard. Patent documents cited in the description • CN 202310492451

[0001] • CN202310820305

[0001]

Claims

1. A cloud platform management method, comprising: grouping a plurality of cloud service components in a cloud platform into a plurality of communities based on call relationships between the plurality of cloud service components, wherein each community comprises at least two of the plurality of cloud service components, a degree of closeness of a call relationship between cloud service components comprised in each community is greater than or equal to a preset degree of closeness, and a degree of closeness between a cloud service component in any one of the plurality of communities and a cloud service component in another one of the plurality of communities is less than the preset degree of closeness; and if a target component in a first community of the plurality of communities meets a first preset condition, creating a second community, wherein the second community is the same as the first community, the second community is configured to share a part of load of the first community, the first preset condition indicates that performance of the target component is lower than preset performance, the first community is any one of the plurality of communities, and the target component is at least one of a plurality of cloud service components in the first community.

2. The method according to claim 1, wherein the call relationship comprises one or more of the following: call information of the plurality of cloud service components in the cloud platform, a call frequency of a link between any two of the plurality of cloud service components in the cloud platform, and an out-degree and an in-degree of any one of the plurality of cloud service components in the cloud platform.

3. The method according to claim 1 or 2, wherein grouping the plurality of cloud service components in the cloud platform into the plurality of communities based on the call relationships between the plurality of cloud service components comprises: grouping the plurality of cloud service components into the plurality of communities based on a community discovery algorithm and the call relationships.

4. The method according to any one of claims 1 to 3, wherein the target component comprises: a cloud service component that meets a second preset condition among the plurality of cloud service components in the first community, or at least one cloud service component specified by a user among the plurality of cloud service components in the first community.

5. The method according to claim 4, wherein the second preset condition comprises top X cloud service components with highest popularity values among the plurality of cloud service components in the first community, wherein the popularity value is a sum of an out-degree of a cloud service component and an in-degree of the cloud service component, and X is an integer greater than 0.

6. The method according to any one of claims 1 to 5, wherein creating the second community if the target component in the first community of the plurality of communities meets the first preset condition comprises: if a first target feature of the target component meets the first preset condition, creating the second community, wherein the first target feature is at least one feature specified by the user among a plurality of features of the target component.

7. The method according to claim 6, wherein the first target feature comprises at least one of CPU usage, memory usage, and an I / O throughput.

8. The method according to claim 7, wherein when the first target feature comprises the CPU usage, the first preset condition comprises: the CPU usage is higher than preset CPU usage; when the first target feature comprises the memory usage, the first preset condition comprises: the memory usage is higher than preset memory usage; or when the first target feature comprises the I / O throughput, the first preset condition comprises: the I / O throughput is higher than a preset I / O throughput.

9. The method according to any one of claims 1 to 8, wherein when creating the second community, the method further comprises: obtaining modularity between the first community and each of other communities than the first community among the plurality of communities, wherein the modularity is a value used to describe a degree of closeness of a call relationship between communities; determining, based on the modularity, whether a target community exists among the other communities, wherein the target community is a community whose modularity with the first community is greater than a threshold among the other communities; and when the target community does not exist among the other communities, creating the second community.

10. The method according to claim 9, wherein the method further comprises: when the target community exists among the other communities, creating a second community combination, wherein the second community combination is the same as a first community combination comprising the first community and the target community, the second community combination comprises the second community and a third community, the third community is the same as the target community, and the second community combination is used to share a part of load of the first community combination.

11. The method according to any one of claims 1 to 10, wherein the method further comprises: if the target component does not meet the first preset condition, determining a hot point and a hot link based on a call relationship between cloud service components in the first community, wherein the hot point is a cloud service component with a relatively high call frequency among the plurality of cloud service components in the first community, and the hot link is a link with a relatively high call frequency between the plurality of cloud service components; and processing the hot point and the hot link based on a preset mitigation policy, wherein the preset mitigation policy is used to enable the processed hot point and hot link to meet respective current load requirements.

12. The method according to claim 11, wherein the hot point is top N cloud service components with highest popularity values among the plurality of cloud service components comprised in the first community, wherein the popularity value is a sum of an out-degree of a cloud service component and an in-degree of the cloud service component, and N is an integer greater than 0; and the hot link is top M links with highest call frequencies among a plurality of links between the plurality of cloud service components comprised in the first community, wherein M is an integer greater than 0.

13. The method according to claim 11 or 12, wherein processing the hot point based on the preset policy comprises: determining a target hot point mitigation policy from a first target correspondence based on an identifier of a community in which the hot point is located and a second target feature of the hot point, wherein the first target correspondence comprises a correspondence between identifiers of the plurality of communities, a plurality of third preset conditions, and a plurality of hot point mitigation policies, the second target feature is a feature specified in advance by the user among a plurality of features of the hot point, and the third preset condition is used to determine whether to perform capacity expansion on the hot point; and processing the hot point based on the target hot point mitigation policy.

14. The method according to claim 13, wherein the hot point mitigation policy comprises: reset scaling and / or horizontal scaling.

15. The method according to any one of claims 11 to 14, wherein processing the hot link based on the preset policy comprises: determining a target hot link mitigation policy from a second target correspondence based on a third target feature of the hot link, wherein the second target correspondence comprises a correspondence between a plurality of fourth preset conditions and a plurality of hot link mitigation policies, and the third target feature is a feature specified in advance by the user among a plurality of features of the hot link; and processing the hot link based on the target hot link mitigation policy.

16. The method according to claim 15, wherein the hot link mitigation policy comprises: lateral scaling and / or longitudinal scaling.

17. The method according to claim 11 or 12, wherein processing the hot point and the hot link based on the preset policy comprises: determining a target type template from a plurality of preset type templates based on a distribution feature of the hot point and the hot link, wherein the distribution feature is used to represent spatial location information of the hot point and the hot link, and the target type template is a type template having a highest degree of matching with the distribution feature among the plurality of preset type templates; determining, from a correspondence between a plurality of type templates and a plurality of mitigation policies, a target mitigation policy corresponding to the target type template; and processing the hot point and the hot link based on the target mitigation policy.

18. A management apparatus, wherein the management apparatus comprises a grouping module and a creation module, wherein the grouping module is configured to group a plurality of cloud service components in a cloud platform into a plurality of communities based on call relationships between the plurality of cloud service components, wherein each community comprises at least two of the plurality of cloud service components, a degree of closeness of a call relationship between cloud service components comprised in each community is greater than or equal to a preset degree of closeness, and a degree of closeness between a cloud service component in any one of the plurality of communities and a cloud service component in another one of the plurality of communities is less than the preset degree of closeness; and the creation module is configured to create a second community if a target component in a first community of the plurality of communities meets a first preset condition, wherein the second community is the same as the first community, the second community is configured to share a part of load of the first community, the first preset condition indicates that performance of the target component is lower than preset performance, the first community is any one of the plurality of communities, and the target component is at least one of a plurality of cloud service components in the first community.

19. The management apparatus according to claim 18, wherein the call relationship comprises one or more of the following: call information of the plurality of cloud service components in the cloud platform, a call frequency of a link between any two of the plurality of cloud service components in the cloud platform, and an out-degree and an in-degree of any one of the plurality of cloud service components in the cloud platform.

20. The management apparatus according to claim 18 or 19, wherein the grouping module is specifically configured to group the plurality of cloud service components into the plurality of communities based on a community discovery algorithm and the call relationships.

21. The management apparatus according to any one of claims 18 to 20, wherein the target component comprises: a cloud service component that meets a second preset condition among the plurality of cloud service components in the first community, or at least one cloud service component specified by a user among the plurality of cloud service components in the first community.

22. The management apparatus according to claim 21, wherein the second preset condition comprises top X cloud service components with highest popularity values among the plurality of cloud service components in the first community, wherein the popularity value is a sum of an out-degree of a cloud service component and an in-degree of the cloud service component, and X is an integer greater than 0.

23. The management apparatus according to any one of claims 18 to 22, wherein the creation module is configured to create the second community if a first target feature of the target component meets the first preset condition, wherein the first target feature is at least one feature specified by the user among a plurality of features of the target component.

24. The management apparatus according to claim 23, wherein the first target feature comprises at least one of processor CPU usage, memory usage, and an I / O throughput.

25. The management apparatus according to claim 24, wherein when the first target feature comprises the CPU usage, the first preset condition comprises: the CPU usage is higher than preset CPU usage; when the first target feature comprises the memory usage, the first preset condition comprises: the memory usage is higher than preset memory usage; or when the first target feature comprises the I / O throughput, the first preset condition comprises: the I / O throughput is higher than a preset I / O throughput.

26. The management apparatus according to any one of claims 18 to 25, wherein the management apparatus further comprises an obtaining module and a determining module, wherein the obtaining module is configured to obtain modularity between the first community and each of other communities than the first community among the plurality of communities, wherein the modularity is a value used to describe a degree of closeness of a call relationship between communities; the determining module is configured to determine, based on the modularity, whether a target community exists among the other communities, wherein the target community is a community whose modularity with the first community is greater than a threshold among the other communities; and the creation module is configured to create the second community when the target community does not exist among the other communities.

27. The management apparatus according to claim 26, wherein the creation module is configured to create a second community combination when the target community exists among the other communities, wherein the second community combination is the same as a first community combination comprising the first community and the target community, the second community combination comprises the second community and a third community, the third community is the same as the target community, and the second community combination is used to share a part of load of the first community combination.

28. The management apparatus according to any one of claims 18 to 27, wherein the determining module is configured to determine a hot point and a hot link based on a call relationship between cloud service components in the first community if the target component does not meet the first preset condition, wherein the hot point is a cloud service component with a relatively high call frequency among the plurality of cloud service components in the first community, and the hot link is a link with a relatively high call frequency between the plurality of cloud service components; and the determining module is further configured to process the hot point and the hot link based on a preset mitigation policy, wherein the preset mitigation policy is used to enable the processed hot point and hot link to meet respective current load requirements.

29. The management apparatus according to claim 28, wherein the hot point is top N cloud service components with highest popularity values among the plurality of cloud service components comprised in the first community, wherein the popularity value is a sum of an out-degree of a cloud service component and an in-degree of the cloud service component, and N is an integer greater than 0; and the hot link is top M links with highest call frequencies among a plurality of links between the plurality of cloud service components comprised in the first community, wherein M is an integer greater than 0.

30. The management apparatus according to claim 27 or 28, wherein the determining module is configured to determine a target hot point mitigation policy from a first target correspondence based on an identifier of a community in which the hot point is located and a second target feature of the hot point, wherein the first target correspondence comprises a correspondence between identifiers of the plurality of communities, a plurality of third preset conditions, and a plurality of hot point mitigation policies, the second target feature is a feature specified in advance by the user among a plurality of features of the hot point, and the third preset condition is used to determine whether to perform capacity expansion on the hot point; and the determining module is further configured to process the hot point based on the target hot point mitigation policy.

31. The management apparatus according to claim 30, wherein the hot point mitigation policy comprises: reset scaling and / or horizontal scaling.

32. The management apparatus according to any one of claims 28 to 31, wherein the determining module is configured to determine a target hot link mitigation policy from a second target correspondence based on a third target feature of the hot link, wherein the second target correspondence comprises a correspondence between a plurality of fourth preset conditions and a plurality of hot link mitigation policies, and the third target feature is a feature specified in advance by the user among a plurality of features of the hot link; and the determining module is further configured to process the hot link based on the target hot link mitigation policy.

33. The management apparatus according to claim 32, wherein the hot link mitigation policy comprises: lateral scaling and / or longitudinal scaling.

34. The management apparatus according to claim 28 or 29, wherein the determining module is configured to determine a target type template from a plurality of preset type templates based on a distribution feature of the hot point and the hot link, wherein the distribution feature is used to represent spatial location information of the hot point and the hot link, and the target type template is a type template having a highest degree of matching with the distribution feature among the plurality of preset type templates; and the determining module is further configured to determine, from a correspondence between a plurality of type templates and a plurality of mitigation policies, a target mitigation policy corresponding to the target type template; and process the hot point and the hot link based on the target mitigation policy.

35. A computing device cluster, comprising at least one computing device, wherein each computing device comprises a processor and a memory; and the processor of the at least one computing device is configured to execute instructions stored in the memory of the at least one computing device, so that the computing device cluster performs the method according to any one of claims 1 to 17.

36. A computer-readable storage medium, storing computer instructions, wherein when the computer instructions are run on a computing device, the computing device is enabled to perform the method according to any one of claims 1 to 17.

37. A computer program product comprising instructions, wherein when the instructions are run by a computing device cluster, the computing device cluster is enabled to perform the method according to any one of claims 1 to 17.

Images