Computing power resource deployment method and device and storage medium

Abstract

The invention provides a computing power resource deployment method and device and a storage medium, relates to the technical field of computers, and can reduce economic cost and improve load balance. The method comprises the following steps: acquiring a deployment request, wherein the deployment request is used for indicating a to-be-deployed target computing power resource; according to the load information and deployment cost of each computing power service platform in the candidate computing power service platforms, deploying target computing power resources; wherein the load information of the computing power service platform is used for reflecting the global load balancing condition after the computing power service platform simulates the deployment of the target computing power resource, and the deployment cost is used for reflecting the cost of the computing power service platform under the condition of deploying the target computing power resource.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a computing resource deployment method, device and storage medium. Background Art

[0002] As digital transformation deepens, the importance of computing networks in economic development becomes increasingly prominent. With the rapid development of technologies such as cloud computing, edge computing, and artificial intelligence, computing demand is growing, and the efficient deployment of computing resources has become the key.

[0003] At present, the research on computing resource deployment methods is mainly in mobile communication scenarios and data center scenarios. For example, effective resource allocation in a communication and positioning integrated network can meet the requirements of communication quality and positioning accuracy at the same time. However, the above method has high economic cost and poor load balancing. Summary of the invention

[0004] The present application provides a computing resource deployment method, device and storage medium, which can reduce economic costs and improve load balancing.

[0005] In order to achieve the above objectives, this application adopts the following technical solutions:

[0006] In a first aspect, the present application provides a computing power resource deployment method, the method comprising: obtaining a deployment request, the deployment request being used to indicate a target computing power resource to be deployed; deploying the target computing power resource according to the load information and deployment cost of each computing power service platform in the candidate computing power service platform; wherein the load information of the computing power service platform is used to reflect the global load balancing situation after the computing power service platform simulates the deployment of the target computing power resources, and the deployment cost is used to reflect the cost of the computing power service platform when deploying the target computing power resources.

[0007] In one possible implementation, target computing resources are deployed according to the load information and deployment cost of each computing service platform among the candidate computing service platforms, including: determining the target computing service platform according to the load information and deployment cost of each computing service platform; and deploying the target computing resources on the target computing service platform.

[0008] In one possible implementation, the method also includes: obtaining, for each computing power service platform, the current resource usage of the computing power service platform; determining, based on the current resource usage of the computing power service platform, the predicted load rate of the computing power service platform after simulating the deployment of target computing power resources; and determining, based on the predicted load rate of each computing power service platform, the load information of the computing power service platform.

[0009] In one possible implementation, the predicted load rate of the computing power service platform after simulating the deployment of target computing power resources is determined based on the current resource usage of the computing power service platform, including: determining the relationship between the proportion of the current resource usage of the computing power service platform in the total resources after simulating the deployment of target computing power resources and a preset load threshold; determining the predicted load rate of the computing power service platform after simulating the deployment of target computing power resources based on the relationship between the current resource usage of the computing power service platform in the total resources after simulating the deployment of target computing power resources and the preset load threshold.

[0010] In one possible implementation, the method also includes: for each computing power service platform, determining the predicted bandwidth resource cost of the computing power service platform after simulating the deployment of target computing power resources based on the usage cost and construction cost of the bandwidth resources of the computing power service platform; determining the deployment cost of the computing power service platform based on the predicted bandwidth resource cost and predicted computing power resource cost of the computing power service platform, the predicted computing power resource cost being the usage cost of the computing power service platform after simulating the deployment of the target computing power resources.

[0011] In one possible implementation, the method also includes: determining the resource concentration of the computing power service platform, the resource concentration is used to reflect the concentration of computing resources after the computing power service platform simulates the deployment of target computing power resources; determining the target computing power service platform based on the load information and deployment cost of each computing power service platform, including: determining the target computing power service platform based on the load information, deployment cost and resource concentration of each computing power service platform.

[0012] In one possible implementation, a target computing power service platform is determined based on the load information, deployment cost, and resource concentration of each computing power service platform, including: determining at least one computing power service platform among the candidate computing power service platforms based on the load information, deployment cost, and resource concentration of each computing power service platform, and at least one computing power service platform meets the target computing power resource deployment requirements; determining the target computing power service platform from at least one computing power service platform based on the deployment priority corresponding to the target computing power resource, and the deployment priority includes the priority corresponding to the load information, the priority corresponding to the deployment cost, and the priority corresponding to the resource concentration.

[0013] In a second aspect, the present application provides a computing power resource deployment device, which includes: a communication unit and a processing unit; the communication unit is used to obtain a deployment request, the deployment request is used to indicate the target computing power resources to be deployed; the processing unit is used to deploy the target computing power resources according to the load information and deployment cost of each computing power service platform in the candidate computing power service platform; wherein the load information of the computing power service platform is used to reflect the global load balancing situation after the computing power service platform simulates the deployment of the target computing power resources, and the deployment cost is used to reflect the cost of the computing power service platform when deploying the target computing power resources.

[0014] In one possible implementation, the processing unit is further used to determine the target computing power service platform based on the load information and deployment cost of each computing power service platform; the processing unit is further used to deploy the target computing power resources on the target computing power service platform.

[0015] In one possible implementation, the communication unit is also used to obtain the current resource usage of the computing power service platform for each computing power service platform; the processing unit is also used to determine the predicted load rate of the computing power service platform after simulating the deployment of target computing power resources based on the current resource usage of the computing power service platform; the processing unit is also used to determine the load information of the computing power service platform based on the predicted load rate of each computing power service platform.

[0016] In one possible implementation, the processing unit is also used to determine the relationship between the current resource usage of the computing power service platform in the total amount of resources after simulating the deployment of the target computing power resources and a preset load threshold; the processing unit is also used to determine the predicted load rate of the computing power service platform after simulating the deployment of the target computing power resources based on the relationship between the current resource usage of the computing power service platform in the total amount of resources after simulating the deployment of the target computing power resources and the preset load threshold.

[0017] In one possible implementation, the processing unit is also used to determine, for each computing power service platform, the predicted bandwidth resource cost of the computing power service platform after simulating the deployment of target computing power resources based on the usage cost and construction cost of the bandwidth resources of the computing power service platform; the processing unit is also used to determine the deployment cost of the computing power service platform based on the predicted bandwidth resource cost and the predicted computing power resource cost of the computing power service platform, and the predicted computing power resource cost is the usage cost of the computing power service platform after simulating the deployment of the target computing power resources.

[0018] In one possible implementation, the processing unit is also used to determine the resource concentration of the computing power service platform, and the resource concentration is used to reflect the concentration degree of computing power resources after the computing power service platform simulates the deployment of target computing power resources; the processing unit is also used to determine the target computing power service platform based on the load information, deployment cost and resource concentration of each computing power service platform.

[0019] In one possible implementation, the processing unit is further used to determine at least one computing power service platform among the candidate computing power service platforms based on the load information, deployment cost, and resource concentration of each computing power service platform, and at least one computing power service platform meets the target computing power resource deployment requirements; the processing unit is also used to determine the target computing power service platform from at least one computing power service platform based on the deployment priority corresponding to the target computing power resource, and the deployment priority includes the priority corresponding to the load information, the priority corresponding to the deployment cost, and the priority corresponding to the resource concentration.

[0020] In a third aspect, the present application provides a computing power resource deployment device, which includes: a processor and a communication interface; the communication interface and the processor are coupled, and the processor is used to run a computer program or instructions to implement the computing power resource deployment method described in the first aspect and any possible implementation method of the first aspect.

[0021] In a fourth aspect, the present application provides a computer-readable storage medium, which stores instructions. When the instructions are executed on a terminal, the terminal executes the computing resource deployment method described in the first aspect and any possible implementation of the first aspect.

[0022] In a fifth aspect, the present application provides a computer program product comprising instructions. When the computer program product is run on a computing power resource deployment device, the computing power resource deployment device executes the computing power resource deployment method described in the first aspect and any possible implementation method of the first aspect.

[0023] In a sixth aspect, the present application provides a chip, comprising a processor and a communication interface, wherein the communication interface and the processor are coupled, and the processor is used to run a computer program or instructions to implement the computing resource deployment method as described in the first aspect and any possible implementation method of the first aspect.

[0024] Specifically, the chip provided in the present application also includes a memory for storing computer programs or instructions.

[0025] The above technical solution brings at least the following beneficial effects: obtaining a deployment request, and deploying the target computing resources according to the load information and deployment cost of each computing service platform in the candidate computing service platform. In other words, the computing resource deployment method provided in the embodiment of the present application takes into account load balancing and deployment cost, thereby reducing economic costs and improving load balancing. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 A structural diagram of a computing power network provided in an embodiment of the present application;

[0027] Figure 2 A schematic diagram of the structure of a communication system provided in an embodiment of the present application;

[0028] Figure 3 A schematic diagram of the composition of a computing resource deployment device provided in an embodiment of the present application;

[0029] Figure 4 A flowchart of a computing resource deployment method provided in an embodiment of the present application;

[0030] Figure 5A flowchart of another computing resource deployment method provided in an embodiment of the present application;

[0031] Figure 6 A flowchart of another computing resource deployment method provided in an embodiment of the present application;

[0032] Figure 7 A flowchart of another computing resource deployment method provided in an embodiment of the present application;

[0033] Figure 8 A flowchart of another computing resource deployment method provided in an embodiment of the present application;

[0034] Fig. 9 A flowchart of another computing resource deployment method provided in an embodiment of the present application;

[0035] Fig.10 A flowchart of another computing resource deployment method provided in an embodiment of the present application;

[0036] Fig.11 A schematic diagram of the structure of a computing resource deployment device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0037] The computing resource deployment method, device and storage medium provided in the embodiments of the present application are described in detail below in conjunction with the accompanying drawings.

[0038] The term "and / or" in this article is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone.

[0039] The terms "first" and "second" and the like in the specification and drawings of this application are used to distinguish different objects, or to distinguish different processing of the same object, rather than to describe a specific order of objects.

[0040] In addition, the terms "including" and "having" and any variations thereof mentioned in the description of the present application are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or devices.

[0041] It should be noted that, in the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0042] In the description of the present application, unless otherwise specified, “plurality” means two or more.

[0043] With the deepening of digital transformation, the importance of computing power network in economic development has become increasingly prominent. Computing power network refers to a new framework that effectively allocates basic resources such as computing and storage between cloud, edge and end through network means on the basis of the continuous ubiquitous development of computing power, and collaboratively provides digital technology and application integration services. Due to the rapid development of technologies such as cloud computing, edge computing and artificial intelligence, computing demand is growing, so the efficient deployment of computing power resources has become the key.

[0044] Figure 1 This is the structural diagram of the computing power network. Figure 1 As shown in the figure, the computing network includes the infrastructure layer, cloud network access layer, computing network perception layer, orchestration and scheduling layer, and service layer.

[0045] The infrastructure layer includes network resources, edge cloud, and central cloud, such as computing nodes, storage devices, network devices, and other hardware resources. The infrastructure layer is the physical foundation of the entire computing network, used to provide computing, storage, and network connectivity capabilities.

[0046] The cloud network access layer includes network resource access and cloud resource access. The cloud network access layer is responsible for handling the interaction between different computing resources and the network, and realizing seamless connection between cloud resources and the network.

[0047] The computing network perception layer includes computing power perception, network perception, application perception, and application programming interface (API) perception. The computing network perception layer is used to collect and analyze network and computing resource status information, monitor key indicators such as network traffic, latency, and packet loss rate, and perceive the load of computing resources, such as central processing unit (CPU) utilization, memory usage, etc.

[0048] The orchestration and scheduling layer includes algorithm management, orchestration management, and task scheduling. The orchestration and scheduling layer is responsible for managing and scheduling computing resources and network resources, automatically allocating appropriate computing and network resources according to application requirements, and realizing dynamic scheduling of tasks to balance loads and improve resource utilization.

[0049] The service layer includes computing power as a service, code as a service, and model as a service. The service layer provides computing services to end users.

[0050] At present, the research on computing resource deployment methods mainly focuses on how to efficiently allocate and schedule computing resources. The following is the key information of the research on computing resource deployment methods.

[0051] 1. Computing and network integration: Through the network distribution service node computing power information, storage information and algorithm information, combined with network information (such as path, latency, etc.), provide the best resource allocation and network connection solutions for user needs. The final form of computing and network integration is expected to form a variety of computing power trading platforms to achieve the optimal use of the entire network resources.

[0052] 2. Unified scheduling: Shanghai has promoted the unified scheduling of computing resources by deploying nine major tasks, which marks that computing power investment has entered a golden period. Large-scale intelligent computing power infrastructure not only provides a prerequisite for technology companies to train large artificial intelligence (AI) models, but also serves as the basis for realizing computing power and algorithm rental services.

[0053] 3. Orchestration algorithm: Innovative research and development of core technologies such as computing power routing algorithm and computing power orchestration algorithm to achieve deep integration of cloud computing and network technology, unified resource management and scheduling. Through dynamic scheduling of computing power resources and automatic allocation of tasks, load is balanced and resource utilization is improved.

[0054] However, the above computing resource deployment method is mainly used in mobile communication scenarios and data center scenarios, with high economic costs and poor load balancing.

[0055] In view of this, the embodiment of the present application provides a computing power resource deployment method, in which the scheduling center obtains a deployment request and deploys the target computing power resource according to the load information and deployment cost of each computing power service platform in the candidate computing power service platform. In other words, the computing power resource deployment method provided by the embodiment of the present application takes into account load balancing and deployment cost, thereby reducing economic costs and improving load balancing.

[0056] The computing resource deployment method provided in the embodiment of the present application is applied to the scheduling layer of the computing network, and is applicable to the problem of computing resource deployment in multi-cloud scenarios. Multi-cloud scenarios involve the collaborative work of public cloud service providers, and different cloud service providers may provide services with different performance at different prices. The computing resource deployment method provided in the embodiment of the present application can select the most suitable computing service platform for resource deployment according to specific resource types, usage costs, and workloads.

[0057] For example, Figure 2A schematic diagram of the structure of a communication system 20 provided in an embodiment of the present application is shown in FIG. The communication system 20 may include at least one resource scheduler 201 and multiple computing service platforms 202 , and the resource scheduler 201 may be connected to the computing service platform 202 for communication. Figure 2 Only one resource scheduler 201 and three computing service platforms 202 are shown. The embodiment of the present application does not impose any restriction on the number of resource schedulers 201 and computing service platforms 202.

[0058] The resource scheduler 201 is used to obtain a deployment request and deploy target computing resources according to the load information and deployment cost of each computing service platform in the candidate computing service platforms.

[0059] Optionally, the resource scheduler 201 may be composed of a server, a switch or a router. The resource scheduler 201 can arrange, schedule and monitor computing resources.

[0060] The computing power service platform 202 is used to provide a large amount of computing and data processing capabilities.

[0061] Optionally, the computing power service platform 202 may be composed of servers, storage devices, and network devices.

[0062] It should be noted that Figure 2 This is just an exemplary framework diagram. Figure 2 The number of nodes included in the Figure 2 In addition to the functional nodes shown, the communication system 20 may also include other nodes, and this application does not impose any limitation on this.

[0063] The application scenarios of the embodiments of the present application are not limited. The system architecture and business scenarios described in the embodiments of the present application are intended to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application. It is known to those skilled in the art that with the evolution of the network architecture and the emergence of new business scenarios, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

[0064] When implementing it specifically, Figure 2 The equipment in Figure 3 The structure shown, or including Figure 3 Parts shown. Figure 3 The present invention provides a schematic diagram of a computing resource deployment device 30, which may be a resource scheduler 201 or a chip or system on chip in the resource scheduler 201. Figure 3 As shown, the computing resource deployment device 30 may include a processor 301 and a bus 302.

[0065] Furthermore, the computing resource deployment device 30 may also include a communication interface 303 and a memory 304. The processor 301, the memory 304 and the communication interface 303 may be connected via a bus 302.

[0066] The processor 301 is a CPU, a general processor, a network processor (NP), a digital signal processor (DSP), a microprocessor, a microcontroller, a programmable logic device (PLD), or any combination thereof. The processor 301 may also be other devices with processing functions, such as circuits, devices, or software modules, without limitation.

[0067] Bus 302 is used to transmit information between the various components included in the computing resource deployment device 30.

[0068] The communication interface 303 is used to communicate with other devices or other communication networks. The other communication networks may be Ethernet, radio access network (RAN), wireless local area network (WLAN), etc. The communication interface 303 may be a module, a circuit, a communication interface or any device capable of achieving communication.

[0069] The memory 304 is used to store instructions, where the instructions may be computer programs.

[0070] The memory 304 may be a read-only memory (ROM) or other types of static storage devices that can store static information and / or instructions, or a random access memory (RAM) or other types of dynamic storage devices that can store information and / or instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), magnetic disk storage media or other magnetic storage devices, etc., without limitation.

[0071] It should be noted that the memory 304 can exist independently of the processor 301, or it can be integrated with the processor 301. The memory 304 can be used to store instructions or program codes or some data, etc. The memory 304 can be located in the computing resource deployment device 30, or it can be located outside the computing resource deployment device 30, without limitation. The processor 301 is used to execute the instructions stored in the memory 304 to implement the computing resource deployment method provided in the following embodiments of the present application.

[0072] In an example, the processor 301 may include one or more CPUs, for example, CPU0 and CPU1 (not shown in the figure).

[0073] As an optional implementation, the computing resource deployment device 30 includes multiple processors.

[0074] As an optional implementation, the computing resource deployment device 30 also includes an output device and an input device. Exemplarily, the input device is a keyboard, a mouse, a microphone, a joystick, and the like, and the output device is a display screen, a speaker, and the like.

[0075] It should be noted that the computing resource deployment device 30 can be a desktop computer, a portable computer, a network server, a mobile phone, a tablet computer, a wireless terminal, an embedded device, a chip system or a Figure 3 In addition, Figure 3 The composition shown in the Figure 2 as well as Figure 3 The limitations of each device in Figure 3 In addition to the parts shown, Figure 2 as well as Figure 3 The various devices in the drawings may include more or fewer components than shown, or combine certain components, or arrange the components differently.

[0076] In the embodiment of the present application, the chip system may be composed of a chip, or may include a chip and other discrete devices.

[0077] In addition, the actions, terms, etc. involved in the various embodiments of the present application can refer to each other without limitation. The message name or parameter name in the message exchanged between the various devices in the embodiments of the present application is only an example, and other names can also be used in the specific implementation without limitation.

[0078] The following describes the computing resource deployment method provided by the embodiment of the present application in conjunction with the accompanying drawings. Among them, the actions, terms, etc. involved in the various embodiments of the present application can refer to each other without limitation. The message name or parameter name in the message exchanged between the various devices in the embodiment of the present application is only an example, and other names can also be used in the specific implementation without limitation. The actions involved in the various embodiments of the present application are only an example, and other names can also be used in the specific implementation, such as: "included in" in the embodiment of the present application can also be replaced by "carried on" or "carried in", etc.

[0079] like Figure 4 As shown, the embodiment of the present application proposes a computing resource deployment method, the method comprising:

[0080] S401: The resource scheduler obtains a deployment request.

[0081] The deployment request is used to indicate the target computing resources to be deployed. The target computing resources include CPU resources, graphics processing unit (GPU) resources, and memory resources.

[0082] In a possible implementation, the user terminal sends a deployment request to the resource scheduler in response to a user instruction. Correspondingly, the resource scheduler receives the deployment request from the user terminal.

[0083] Exemplarily, the above deployment request can be represented by u, which has the following attribute set C u .

[0084] C u ={C c,u ,C g,u ,C r,u}

[0085] Among them, C u represents the target computing resources to be deployed, C c,u Indicates CPU resources, C g,u Indicates GPU resources, C r,u Represents memory resources. Specifically, CPU resources are the ratio of the number of CPU cores to the number of virtual CPU (vCPU) cores, in units. GPU resources are the ratio of the number of GPU cards to the number of virtual GPU (vGPU) cards, in units. Memory resources are the actual storage capacity, in gigabytes (GB). Bandwidth resources are the ability to transmit data, in gigabits per second (Gbps).

[0086] S402: The resource scheduler deploys target computing resources according to the load information and deployment cost of each computing service platform in the candidate computing service platforms.

[0087] The load information of the computing power service platform is used to reflect the global load balancing situation after the computing power service platform simulates the deployment of the target computing power resources. The deployment cost is used to reflect the cost of the computing power service platform when deploying the target computing power resources.

[0088] In one possible implementation, the resource scheduler determines the target computing service platform based on the load information and deployment cost of each computing service platform, and deploys the target computing resources on the target computing service platform. Figure 4 The described embodiments are not described in detail here.

[0089] Exemplarily, the load information may be a value of a standard deviation of the load rate. The smaller the load information, the more balanced the load.

[0090] For example, the set of candidate computing service platforms can be represented as M={1, 2, ..., m}, where m is a positive integer and represents the number of the computing service platform. The computing resources in the candidate computing service platform numbered i can be represented as C i ={C c,i ,C g,i ,C r,i The delay between the candidate computing service platform numbered i and the candidate computing service platform numbered j can be expressed as T i,j , the unit is millisecond (ms).

[0091] It should be noted that the above-mentioned target computing service platform can be an existing computing service platform, and the target computing resources can be deployed in the existing computing service platform as a resource expansion of the existing computing service platform. Alternatively, a new computing service platform can be built as the target computing service platform, and the target computing resources can be deployed in the new computing service platform.

[0092] Furthermore, the resource scheduler can also set up multiple plans to deploy the target computing resources to multiple existing or newly built computing service platforms, including the splitting method of the target computing resources and the computing service platform to which the split computing resources are deployed. For each deployment plan, the resource scheduler can calculate the load information and deployment cost corresponding to the deployment plan, and determine the final deployment plan with the optimization goal of minimizing the load information and deployment cost.

[0093] In the computing resource deployment method provided by the present application, the dispatch center obtains the deployment request and deploys the target computing resource according to the load information and deployment cost of each computing service platform in the candidate computing service platform. In other words, the computing resource deployment method provided by the embodiment of the present application takes into account load balancing and deployment cost, thereby reducing economic costs and improving load balancing.

[0094] In one embodiment, Figure 5 As shown, the above S402 can be specifically determined through the following S501 to S502.

[0095] S501. The resource scheduler determines the target computing service platform based on the load information and deployment cost of each computing service platform.

[0096] In one possible implementation, if the computing power service platform corresponding to the minimum load information and the computing power service platform corresponding to the minimum deployment cost are the same computing power service platform, the dispatching center will use the computing power service platform as the target computing power service platform.

[0097] In another possible implementation method, if the computing power service platform corresponding to the minimum load information and the computing power service platform corresponding to the minimum deployment cost are not the same computing power service platform, the dispatching center selects the corresponding computing power service platform as the target computing power service platform based on the importance of the load information and the deployment cost, or selects other computing power service platforms with load information and deployment costs within the allowable range as the target computing power service platform.

[0098] It should be noted that the above-mentioned target computing power service platform can be an existing computing power service platform or a newly built computing power service platform.

[0099] S502: The resource scheduler deploys target computing resources on the target computing service platform.

[0100] In one possible implementation, the resource scheduler sends a deployment work order to the relevant personnel. In response, the relevant personnel receive the deployment work order from the resource scheduler and deploy computing resources on the target computing service platform according to the deployment work order. The deployment work order includes a specific hardware device installation plan. After the hardware device is installed, the resource scheduler registers and configures the computing resources, connects the computing resources to the target computing service platform, and performs performance tests on the computing resources to ensure that the computing resources can be correctly identified and used.

[0101] Furthermore, before S402, the resource scheduler needs to determine the load information of the computing power service platform so that the target computing power service platform can be determined based on the load information. Figure 6 As shown, the computing resource deployment method recorded in the embodiment of the present application may also include the following steps.

[0102] S601. For each computing power service platform, the resource scheduler obtains the current resource usage of the computing power service platform.

[0103] Among them, the current resource usage of the computing power service platform includes the used CPU resources, used GPU resources, and used memory resources.

[0104] Exemplarily, the current resource usage of the computing power service platform can be represented by the following set.

[0105] C i,used ={C c,i,used ,C g,i,used ,C r,i,used}

[0106] Among them, C i,used Indicates the current resource usage of the computing service platform numbered i, C c,i,used Indicates the CPU resources currently used by the computing service platform numbered i, C g,i,used Indicates the GPU resources currently used by the computing service platform numbered i, C r,i,used Indicates the memory resources currently used by the computing power service platform numbered i.

[0107] S602. The dispatch center determines the predicted load rate of the computing power service platform after simulating the deployment of target computing power resources based on the current resource usage of the computing power service platform.

[0108] In one possible implementation, the resource scheduler determines the CPU resource load rate of the computing power service platform based on the CPU resources currently used by the computing power service platform, the CPU resource load threshold, and the weight corresponding to the CPU resources. Similarly, the resource scheduler uses the same method to determine the GPU resource load rate and memory resource load rate of the computing power service platform, and determines the predicted load rate of the computing power service platform based on the CPU resource load rate, GPU resource load rate, and memory resource load rate. For details, please refer to the following Figure 6 The described embodiments are not described in detail here.

[0109] S603. The resource scheduler determines the load information of the computing power service platform according to the predicted load rate of each computing power service platform.

[0110] In one possible implementation method, the dispatch center calculates the average load rate of all computing power service platforms based on the predicted load rate of each computing power service platform, calculates the standard deviation of the load rate based on the average load rate, and determines the standard deviation of the load rate as the load information of the computing power service platform.

[0111] The average value of the above load factor can be expressed by the following formula 1.

[0112]

[0113] Among them, AL represents the average value of the load rate, Load n It indicates the predicted load rate of computing service platform numbered n, and m indicates the number of computing service platforms.

[0114] The standard deviation of the above load factor can be expressed by the following formula 2.

[0115]

[0116] Where V represents the standard deviation of the load factor, Load n It represents the predicted load rate of the computing service platform numbered n, AL represents the average load rate, and m represents the number of computing service platforms.

[0117] In one embodiment, Figure 7 As shown, the above S602 can be specifically determined through the following S701 to S702.

[0118] S701. The resource scheduler determines the relationship between the current resource usage of the computing power service platform and the preset load threshold as a proportion of the total resources after simulating the deployment of the target computing power resources.

[0119] Exemplarily, the above-mentioned preset load thresholds may include a CPU resource load threshold, a GPU resource load threshold, and a memory resource load threshold.

[0120] In combination with the above example, the implementation process of the above S701 can be: the resource scheduler calculates the proportion of the CPU resources currently used by the computing power service platform in the total CPU resources after simulating the deployment of the target computing power resources, and compares the proportion with the CPU resource load threshold to determine the size relationship between the CPU resource proportion and the CPU resource load threshold. Similarly, the resource scheduler uses the same method to determine the size relationship between the GPU resource proportion and the GPU resource load threshold and the size relationship between the memory resource proportion and the memory resource load threshold.

[0121] S702. The resource scheduler determines the predicted load rate of the computing power service platform after simulating the deployment of the target computing power resources based on the relationship between the current resource usage of the computing power service platform in the total resources after simulating the deployment of the target computing power resources and the preset load threshold.

[0122] Exemplarily, the predicted load rate includes a CPU resource load rate, a GPU resource load rate, and a memory resource load rate, which can be expressed by the following formula 3.

[0123] Load i = Loadi,c +Load i,g +Load i,r Formula 3

[0124] Among them, Load i Indicates the predicted load rate of the computing service platform numbered i, Load i,c Indicates the CPU resource load rate of the computing power service platform numbered i, Load i,g Indicates the GPU resource load rate of the computing service platform numbered i, Load i,r Indicates the memory resource load rate of the computing power service platform numbered i.

[0125] The following specifically describes the calculation methods of the CPU resource load rate, GPU resource load rate, and memory resource load rate.

[0126] 1-1. CPU resource load rate

[0127] The CPU resource load rate can be expressed by the following formula 4.

[0128]

[0129] Among them, Load i,c Indicates the CPU resource load rate of the computing service platform numbered i, C c,i,used Indicates the CPU resources currently used by the computing service platform numbered i, g cpu Indicates the CPU resource load threshold, C c,i Indicates the total amount of CPU resources before the target computing resources are deployed on the computing service platform numbered i. c,u Indicates the CPU resources to be deployed. a1 indicates that the CPU resources do not exceed the CPU resource load threshold g. cpu The weight at that time, k cpu Indicates that the CPU resource exceeds the CPU resource load threshold g cpu The weight after that.

[0130] 1-2. GPU resource load rate

[0131] The GPU resource load rate can be expressed by the following formula 5.

[0132]

[0133] Among them, Load i,g Indicates the GPU resource load rate of the computing service platform numbered i, C g,i,used Indicates the GPU resources currently used by the computing service platform numbered i, g gpu Indicates the GPU resource load threshold, C g,iIndicates the total amount of GPU resources before the target computing resources are deployed on the computing service platform numbered i. g,u Indicates the GPU resources to be deployed. a2 indicates that the GPU resources do not exceed the GPU resource load threshold g. gpu The weight at that time, k gpu Indicates that the GPU resources exceed the GPU resource load threshold g gpu The weight after that.

[0134] 1-3. Memory resource load rate

[0135] The memory resource load rate can be expressed by the following formula 6.

[0136]

[0137] Among them, Load i,r Indicates the memory resource load rate of the computing service platform numbered i, C r,i,used Indicates the memory resources currently used by the computing service platform numbered i, g ram Indicates the memory resource load threshold, C r,i Indicates the total amount of memory resources of the computing service platform numbered i before the target computing resources are deployed, C r,u Indicates the memory resources to be deployed. a3 indicates that the memory resources do not exceed the memory resource load threshold g. ram The weight at that time, k ram Indicates that the memory resource exceeds the memory resource load threshold g ram The weight after that.

[0138] It should be noted that a1+a2+a3=1. If the target computing power resources are used for CPU-intensive tasks, the values ​​of a1, a2 and a3 can be 0.6, 0.1 and 0.3 respectively. If the target computing power resources are used for GPU acceleration tasks, the values ​​of a1, a2 and a3 can be 0.3, 0.5 and 0.2 respectively. If the target computing power resources are used for memory-intensive tasks, the values ​​of a1, a2 and a3 can be 0.2, 0.1 and 0.7 respectively.

[0139] The above cpu The value of g can be between 70% and 80%. gpu The value of g can be between 80% and 90%. ram The value of can be between 80% and 85%. cpu , g gpu and ram The values ​​of can be 75%, 85% and 80% respectively.

[0140] The above k cpu , k gpu and kram The value of can be 10 times of a1, a2 and a3 respectively.

[0141] Furthermore, before S402, the dispatch center needs to determine the deployment cost of the computing power service platform so that the target computing power service platform can be determined based on the deployment cost. Figure 8 As shown, the computing resource deployment method recorded in the embodiment of the present application may also include the following steps.

[0142] S801. For each computing power service platform, the resource scheduler determines the predicted bandwidth resource cost of the computing power service platform after simulating the deployment of target computing power resources based on the usage cost and construction cost of the bandwidth resources of the computing power service platform.

[0143] It should be noted that if the available bandwidth resources of the computing power service platform meet the bandwidth resource requirements of the target computing power resources, the predicted bandwidth resource cost of the computing power service platform after simulating the deployment of the target computing power resources is the usage cost of the bandwidth resources of the computing power service platform. If the available bandwidth resources of the computing power service platform do not meet the bandwidth resource requirements of the target computing power resources, the predicted bandwidth resource cost of the computing power service platform after simulating the deployment of the target computing power resources is the usage cost and construction cost of the bandwidth resources of the computing power service platform.

[0144] Exemplarily, the above-mentioned predicted bandwidth resource cost can be expressed by the following formula 7.

[0145] P i,u1 =B u1 P i,b1 +(B u -B u1 ) i,b2 Formula 7

[0146] Among them, P i,u1 represents the predicted bandwidth resource cost of the computing service platform numbered i, B u1 represents the bandwidth resources that can be directly used by the computing power service platform, B u represents the bandwidth resource requirement of the target computing resource, B u Can be obtained from the deployment request, B u -B u1 Indicates that the computing service platform needs to create new bandwidth resources, P i,b1 P represents the usage fee of the computing service platform numbered i for the unit bandwidth resource. i,b2 It represents the construction cost of the computing service platform numbered i for unit bandwidth resources.

[0147] S802. The resource scheduler determines the deployment cost of the computing power service platform based on the predicted bandwidth resource cost and the predicted computing power resource cost of the computing power service platform.

[0148] Among them, the predicted computing power resource cost is the usage cost after the computing power service platform simulates the deployment of the target computing power resources.

[0149] In one possible implementation, the dispatch center obtains the economic cost information of the computing service platform, calculates the predicted computing resource cost of the computing service platform, and adds the predicted computing resource cost and predicted bandwidth resource cost of the computing service platform to obtain the deployment cost of the computing service platform. The economic cost information includes the unit price of CPU resources, GPU resources, and memory resources.

[0150] Exemplarily, the deployment cost of the above computing power service platform can be expressed by the following formula 8.

[0151] P i,u =P i,u1 +P i,u2 Formula 8

[0152] Among them, P i,u represents the deployment cost of computing service platform numbered i, P i,u1 represents the predicted bandwidth resource cost of the computing service platform numbered i, P i,u1 It can be calculated from the relevant description in S701 above, P i,u2 Indicates the predicted computing resource cost of the computing service platform numbered i.

[0153] Furthermore, the predicted computing resource cost of the computing service platform can be expressed by the following formula 9.

[0154] P i,u2 =C c,u P i,c +C g,u P i,g +C r,u P i,r Formula 9

[0155] Among them, P i,u2 represents the predicted computing resource cost of the computing service platform numbered i, C c,u , C g,u and C r,u Respectively represent the CPU resources, GPU resources and memory resources to be deployed, P i,c , P i,g and P i,r They respectively represent the charges of the computing service platform numbered i for the use of unit CPU resources, GPU resources, and memory resources.

[0156] It should be noted that in some areas, the operation and maintenance, computer room and electricity costs are low and subsidized, so the computing power resource charges are low, but the bandwidth resources are scarce and need to be newly built or supplemented, so the bandwidth resource charges are high. In other areas, the operation and maintenance, computer room and electricity costs are high, so the computing power resource charges are high, but the bandwidth resources are sufficient, so the bandwidth resource charges are low.

[0157] Furthermore, before S402, the resource scheduler may also determine the resource concentration of the computing power service platform, and then determine the target computing power service platform based on the load information, deployment cost and resource concentration of the computing power service platform. Fig. 9 As shown, the computing resource deployment method recorded in the embodiment of the present application may also include the following steps.

[0158] S901. The dispatch center determines the resource concentration of the computing power service platform.

[0159] Among them, resource concentration is used to reflect the concentration of computing resources after the computing service platform simulates the deployment of target computing resources.

[0160] In one possible implementation, the dispatch center determines the resource concentration of the computing power service platform based on the computing power resources of each computing power service platform, the latency between each computing power service platform, and the weights corresponding to different types of computing power resources.

[0161] Optionally, if the target computing power resources are deployed in a computing power service platform, the resource concentration of the computing power service platform can satisfy the following formula 10.

[0162]

[0163] In the above formula, C j =α1·C c,j +α2·C g,j +α3·C r,j

[0164] C k =α1·C c,k +α2·C g,k +α3·C r,k

[0165] C u =α1·C c,u +α2·C g,u +α3·C r,u

[0166] Among them, E i,u Indicates the resource concentration of computing service platform numbered i, C jrepresents the weighted computing resources of the computing service platform numbered j, T i,j represents the delay between the computing service platform numbered i and the computing service platform numbered j, C u represents the weighted target computing resources to be deployed, C k Represents the weighted computing resources of the computing service platform numbered k. α1, α2, and α3 represent the weights of CPU resources, GPU resources, and memory resources, respectively.

[0167] Alternatively, if the target computing power resources are split and deployed in multiple computing power service platforms, the resource aggregation of the computing power service platform can satisfy the following formula 11 (for example, the set of candidate computing power service platforms is N = {1, 2, ..., n}, C u,i Refers to the computing resources deployed on the computing service platform numbered i, i∈N, C u,j Refers to the computing resources deployed on the computing service platform numbered j. ).

[0168]

[0169] In the above formula,

[0170]

[0171] C k =α1·C c,k +α2·C g,k +α3·C r,k

[0172] C u,i =α1·C c,u,i +α2·C g,u,i +α3·C r,u,i

[0173] Among them, E u represents the resource aggregation degree under the split deployment scheme, C u,i Represents the computing resources deployed on the computing service platform numbered i, i∈N, C u,j represents the computing resources deployed on the computing service platform numbered j, C k represents the weighted computing resources of the computing service platform numbered k, T i,j Indicates the delay between computing service platform numbered i and computing service platform numbered j.

[0174] It is understandable that a good resource aggregation can provide more stable services, faster response speed and better resource management, ensuring that resource distribution is neither too dispersed nor too concentrated, avoiding local overload or waste, and thus improving user experience. The aggregation of a large number of computing resources can also produce economies of scale, improve resource utilization efficiency and computing speed, and reduce costs.

[0175] In addition, from the user's perspective, a small amount of computing power demand usually has high requirements for real-time performance and latency. In this case, computing power resources can be deployed in a large computing power cluster or nearby (for example, close to the user or where the data is generated), thereby reducing data transmission delays and improving response speed. For situations that require a lot of computing power support (for example, large-scale AI model training, high-performance computing, etc.), computing power resources can be deployed in remote areas with weak computing power resources, which can reduce operating costs and reduce pressure on urban infrastructure.

[0176] S902. The resource scheduler determines the target computing service platform based on the load information, deployment cost, and resource concentration of each computing service platform.

[0177] In one possible implementation, the resource scheduler determines the target computing service platform based on the load information, deployment cost, and resource concentration of each computing service platform, combined with the deployment priority. Fig.10 The described embodiments are not described in detail here.

[0178] In another possible implementation, the resource scheduler determines the usage cost of each computing service platform based on the ratio between the deployment cost of each computing service platform and the resource concentration. The resource scheduler center obtains a set of non-dominated solutions (i.e., Pareto solution set) from the usage cost and load information of each computing service platform based on the resource deployment optimization target min (usage cost, load information), and selects the solution in the middle of the Pareto frontier as the target computing service platform.

[0179] It can be understood that a better compromise can be achieved by selecting a solution located in the middle of the Pareto front as the target computing power service platform.

[0180] In one embodiment, Fig.10 As shown, the above S902 can be specifically determined through the following S1001 to S1002.

[0181] S1001. The resource scheduler determines at least one computing power service platform among the candidate computing power service platforms based on the load information, deployment cost, and resource concentration of each computing power service platform.

[0182] Among them, at least one computing power service platform meets the target computing power resource deployment requirements.

[0183] In one possible implementation, the resource scheduler calculates the ratio between the deployment cost and resource concentration of each computing power service platform, determines the usage cost of each computing power service platform, and selects at least one computing power service platform that meets the target computing power resource deployment requirements from the candidate computing power service platforms based on the usage cost and load information of each computing power service platform.

[0184] Exemplarily, the target computing power resource deployment requirement may be that the usage cost of the computing power service platform is less than a cost threshold and / or the load information of the computing power service platform is less than a load threshold.

[0185] S1002. The resource scheduler determines a target computing power service platform from at least one computing power service platform according to the deployment priority corresponding to the target computing power resource.

[0186] The deployment priority includes the priority corresponding to the load information, the priority corresponding to the deployment cost, and the priority corresponding to the resource concentration.

[0187] In one possible implementation, the resource scheduler determines a target computing power service platform from at least one computing power service platform according to the highest priority among the deployment priorities corresponding to the target computing power resources.

[0188] For example, when the priority corresponding to the deployment cost is the highest, the resource scheduler determines the computing power service platform with the smallest deployment cost among at least one computing power service platform as the target computing power service platform.

[0189] It can be understood that the resource scheduler determines the target computing service platform based on the deployment priority corresponding to the target computing resources. This takes into account the importance of load information, deployment cost, and resource concentration, and can select a computing service platform that better meets the needs as the target computing service platform.

[0190] The following are scenarios in which the computing resource deployment method provided in the embodiments of the present application can be applied.

[0191] Scenario 1: Deploy a batch of new computing resources in the computing network. These resources are deployed in a single computing service platform. This computing service platform can be an existing one or a newly built one. The specific implementation steps are as follows.

[0192] Step 1: The resource scheduler determines the addresses where computing power can be deployed globally in the computing network. The set is D. Each address has no computing service platform or has one computing service platform. The resource scheduler obtains each address D in real time. i network resources, computing resources, and economic cost information.

[0193] Step 2: The resource scheduler calculates each address D for the address set D. i The computing power usage cost W under the deployment request i,u .

[0194] Among them, the computing power usage cost W i,u This is equivalent to the ratio between the deployment cost and the resource concentration described in the aforementioned S902.

[0195] Step 3: The resource scheduler calculates each address D for the address set D. i Load balancing standard deviation V under deployment requests i,u .

[0196] Among them, the load balancing standard deviation V i,u Equivalent to the load information described above.

[0197] Step 4: Based on the calculation results of steps 2 and 3, the resource scheduler determines a set of non-dominated solutions, namely, the Pareto solution set.

[0198] Step 5: The resource scheduler takes the solution in the middle of the Pareto front as the target computing power service platform.

[0199] Scenario 2: Deploy a batch of new computing resources in the computing network. These resources are deployed in multiple computing service platforms. These computing service platforms can be existing or newly built. The specific implementation steps are as follows.

[0200] Step 6: The resource scheduler determines the addresses where the computing power can be deployed globally in the computing power network, which are set as D. Each address has no computing power service platform or has a computing power service platform. The resource scheduler obtains each address D in real time. i network resources, computing resources, and economic cost information.

[0201] Step 7: The resource scheduler sets multiple plans for deploying computing resources to multiple computing service platforms. The plans include the splitting method of the computing resources to be deployed and the computing service platforms to which the split computing resources are deployed, forming a set N of split deployment plans.

[0202] Step 8: The resource scheduler calculates the computing power usage cost W of each deployment plan under the deployment request for the set of split deployment plans N. i,u .

[0203] Step 9: The resource scheduler calculates the load balancing standard deviation V of each deployment plan under the deployment request for the split deployment plan set N. i,u .

[0204] Step 10: Based on the calculation results of steps 8 and 9, the resource scheduler determines a set of non-dominated solutions, namely, the Pareto solution set.

[0205] Step 11: The resource scheduler takes the split deployment plan corresponding to the solution located in the middle of the Pareto front as the target deployment plan.

[0206] It is understandable that the above-mentioned computing power resource deployment method can be implemented by a computing power resource deployment device. In order to realize the above-mentioned functions, the computing power resource deployment device includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should easily realize that, in combination with the modules and algorithm steps of each example described in the embodiments disclosed herein, the embodiments disclosed in this application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the embodiments disclosed in this application.

[0207] The embodiments disclosed in this application can divide the functional modules of the computing resource deployment device generated by the above method example. For example, each functional module can be divided corresponding to each function, or two or more functions can be integrated into one processing module. The above integrated modules can be implemented in the form of hardware or in the form of software functional modules. It should be noted that the division of modules in the embodiments disclosed in this application is schematic and is only a logical functional division. There may be other division methods in actual implementation.

[0208] Fig.11 A schematic diagram of a computing resource deployment device provided in an embodiment of the present application. Fig.11 As shown, the computing resource deployment device 110 can be used to execute Figure 4-Figure 10 The computing resource deployment method shown in the figure. The computing resource deployment device 110 includes: a communication unit 1101 and a processing unit 1102.

[0209] The communication unit 1101 is used to obtain a deployment request, which is used to indicate the target computing resources to be deployed; the processing unit 1102 is used to deploy the target computing resources according to the load information and deployment cost of each computing service platform in the candidate computing service platforms; wherein the load information of the computing service platform is used to reflect the global load balancing situation after the computing service platform simulates the deployment of the target computing resources, and the deployment cost is used to reflect the cost of the computing service platform when deploying the target computing resources.

[0210] In one possible implementation, the processing unit 1102 is further used to determine the target computing power service platform based on the load information and deployment cost of each computing power service platform; the processing unit 1102 is further used to deploy the target computing power resources on the target computing power service platform.

[0211] In one possible implementation, the communication unit 1101 is also used to obtain the current resource usage of the computing power service platform for each computing power service platform; the processing unit 1102 is also used to determine the predicted load rate of the computing power service platform after simulating the deployment of target computing power resources based on the current resource usage of the computing power service platform; the processing unit 1102 is also used to determine the load information of the computing power service platform based on the predicted load rate of each computing power service platform.

[0212] In one possible implementation, processing unit 1102 is also used to determine the relationship between the current resource usage of the computing power service platform in the total amount of resources after simulating the deployment of the target computing power resources and a preset load threshold; processing unit 1102 is also used to determine the predicted load rate of the computing power service platform after simulating the deployment of the target computing power resources based on the relationship between the current resource usage of the computing power service platform in the total amount of resources after simulating the deployment of the target computing power resources and the preset load threshold.

[0213] In one possible implementation, processing unit 1102 is also used to determine, for each computing power service platform, the predicted bandwidth resource cost of the computing power service platform after simulating the deployment of target computing power resources based on the usage cost and construction cost of the bandwidth resources of the computing power service platform; processing unit 1102 is also used to determine the deployment cost of the computing power service platform based on the predicted bandwidth resource cost and the predicted computing power resource cost of the computing power service platform, and the predicted computing power resource cost is the usage cost of the computing power service platform after simulating the deployment of the target computing power resources.

[0214] In one possible implementation, processing unit 1102 is also used to determine the resource concentration of the computing power service platform, and the resource concentration is used to reflect the concentration degree of computing power resources after the computing power service platform simulates the deployment of target computing power resources; processing unit 1102 is also used to determine the target computing power service platform based on the load information, deployment cost and resource concentration of each computing power service platform.

[0215] In one possible implementation, processing unit 1102 is further used to determine at least one computing power service platform among the candidate computing power service platforms based on the load information, deployment cost, and resource concentration of each computing power service platform, and at least one computing power service platform meets the target computing power resource deployment requirements; processing unit 1102 is also used to determine the target computing power service platform from at least one computing power service platform based on the deployment priority corresponding to the target computing power resource, and the deployment priority includes the priority corresponding to the load information, the priority corresponding to the deployment cost, and the priority corresponding to the resource concentration.

[0216] Through the description of the above implementation methods, technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device and unit described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0217] The present disclosure also provides a computer-readable storage medium having instructions stored thereon. When the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the computing resource deployment method provided by the above-mentioned embodiment of the present disclosure.

[0218] The embodiments of the present disclosure also provide a computer program product containing instructions, which, when executed on an electronic device, enables the electronic device to execute the computing resource deployment method provided by the embodiments of the present disclosure.

[0219] Among them, the computer readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a register, a hard disk, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above, or any other form of computer readable storage medium known in the art. An exemplary storage medium is coupled to a processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an application-specific integrated circuit (ASIC). In the embodiments of the present application, a computer-readable storage medium may be any tangible medium that contains or stores a program, which may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0220] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A computing resource deployment method, characterized in that: The method comprises: Obtaining a deployment request, where the deployment request is used to indicate a target computing resource to be deployed; The target computing power resources are deployed according to the load information and deployment cost of each computing power service platform among the candidate computing power service platforms; wherein the load information of the computing power service platform is used to reflect the global load balancing situation after the computing power service platform simulates the deployment of the target computing power resources, and the deployment cost is used to reflect the cost of the computing power service platform when deploying the target computing power resources.

2. The method according to claim 1, characterized in that: The deploying the target computing power resources according to the load information and deployment cost of each computing power service platform in the candidate computing power service platforms includes: Determine the target computing service platform according to the load information and deployment cost of each computing service platform; Deploy the target computing power resources on the target computing power service platform.

3. The method according to claim 2, characterized in that The method further comprises: For each computing power service platform, obtaining the current resource usage of the computing power service platform; Determine, based on the current resource usage of the computing power service platform, a predicted load rate of the computing power service platform after simulating the deployment of the target computing power resources; Determine the load information of the computing power service platform according to the predicted load rate of each computing power service platform.

4. The method according to claim 3, characterized in that The determining, according to the current resource usage of the computing power service platform, a predicted load rate of the computing power service platform after simulating the deployment of the target computing power resources, includes: Determine the relationship between the current resource usage of the computing power service platform and the preset load threshold; According to the relationship between the proportion of the current resource usage of the computing power service platform in the total resources after the simulated deployment of the target computing power resources and the preset load threshold, the predicted load rate of the computing power service platform after the simulated deployment of the target computing power resources is determined.

5. The method according to claim 2, characterized in that: The method further comprises: For each computing power service platform, determining a predicted bandwidth resource cost of the computing power service platform after simulating the deployment of the target computing power resources according to the usage cost and construction cost of the bandwidth resources of the computing power service platform; Based on the predicted bandwidth resource cost and predicted computing power resource cost of the computing power service platform, the deployment cost of the computing power service platform is determined, and the predicted computing power resource cost is the usage cost of the computing power service platform after simulating the deployment of the target computing power resources.

6. The method according to any one of claims 2 to 5, characterized in that: The method further comprises: Determine the resource concentration degree of the computing power service platform, where the resource concentration degree is used to reflect the concentration degree of computing power resources after the computing power service platform simulates the deployment of the target computing power resources; The determining the target computing power service platform according to the load information and deployment cost of each computing power service platform includes: The target computing power service platform is determined according to the load information, deployment cost and resource concentration of each computing power service platform.

7. The method according to claim 6, characterized in that The determining the target computing service platform according to the load information, deployment cost and resource concentration of each computing service platform includes: Determine at least one computing service platform among the candidate computing service platforms according to the load information, deployment cost, and resource concentration of each computing service platform, and the at least one computing service platform meets the target computing resource deployment requirements; According to the deployment priority corresponding to the target computing power resources, the target computing power service platform is determined from the at least one computing power service platform, wherein the deployment priority includes the priority corresponding to the load information, the priority corresponding to the deployment cost, and the priority corresponding to the resource concentration.

8. A computing resource deployment device, characterized in that: The device comprises: a communication unit and a processing unit; The communication unit is used to obtain a deployment request, where the deployment request is used to indicate a target computing resource to be deployed; The processing unit is used to deploy the target computing power resources according to the load information and deployment cost of each computing power service platform in the candidate computing power service platforms; wherein the load information of the computing power service platform is used to reflect the global load balancing situation after the computing power service platform simulates the deployment of the target computing power resources, and the deployment cost is used to reflect the cost of the computing power service platform when deploying the target computing power resources.

9. A computing resource deployment device, characterized in that: include: Memory and processor; The memory is coupled to the processor; The memory is used to store instructions executable by the processor; When the processor executes the instruction, it performs the computing resource deployment method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and when the computer instructions are executed on a computer, the computer executes the computing resource deployment method as described in any one of claims 1 to 7.

11. A computer program product, characterized in that The computer program product includes computer program instructions, and when the computer program instructions are executed by a processor, the computing power resource deployment method as described in any one of claims 1 to 7 is implemented.

Images