Method and system for balancing cloud service resources of server

Abstract

The invention discloses a method and a system for realizing server cloud service resource equalization. The method comprises the following steps: presetting and storing a constraint relationship between performance configuration purchased by a user and PDP configuration; counting the occupancy rate of the performance resource currently used by the user in the performance configuration purchased by the user; judging whether the occupancy rate exceeds a set threshold value or not, and if yes, performing the next step; reading the performance configuration purchased by the current user, and obtaining the PDP configuration corresponding to the performance configuration of the user according to the constraint relationship; and judging whether the current PDP configuration of the user is the same as the corresponding PDP configuration or not, if not, reducing one level of PDP configuration on the basis of the current PDP configuration, or reducing the current PDP configuration to the corresponding PDP configuration, and configuring all CPUs allocated by the current user according to the reduced PDP configuration. Compared with the prior art, the cloud service fairness is kept while the cloud resources are effectively utilized, and the user experience with high purchase configuration is protected.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of cloud computing, and particularly relates to a method and system for realizing server cloud service resource balancing. BACKGROUND

[0002] In the field of cloud computing, users can purchase different servers in different ways. For an arm server, assuming that there are 160 cores, the server can be rented and sold to users in core units, such as 2G per core, 8G per dual-core, 16G per quad-core, and the like. The server allocates these resources to users.

[0003] With the increase in the number of purchases, a server can have up to hundreds of users, some of whom purchase the server to build a website, some to store data, some to play games, and some to process AI data. Generally, a fixed physical CPU is allocated to the user, but the SLC and DDR on the SOC are common resources for all CPUs. Some users who consume more resources will inevitably affect the use of other users, because the SLC and DDR are common resources and must be accessed by all users. Therefore, when the number of users is too large, the CPU and memory resources of the server need to be limited according to the demand of the user and the amount of resources.

[0004] However, for some users who purchase high CPU core numbers and high memory resources, the performance of the cloud service will be unnecessarily limited due to the current low resource demand, resulting in a poor experience, and it is difficult to balance fairness and user experience.

[0005] Therefore, there is an urgent need for a service resource balancing method and system that can solve the above problems. SUMMARY

[0006] The present application aims to provide a method and system for realizing server cloud service resource balancing, which can reduce the PDP configuration of a user or make the PDP configuration of the user reach the corresponding minimum PDP configuration according to the performance configuration purchased by the user when the user has a high occupancy rate, thereby effectively utilizing cloud resources while maintaining the fairness of cloud services and protecting the experience of users who purchase high performance configurations.

[0007] In order to achieve the above object, the application provides a method for realizing server cloud service resource balance, comprising: setting and storing in advance a constraint relationship between performance configuration purchased by a user and PDP configuration; dividing the performance configuration into multiple gears according to the size of the performance configuration purchased by the user, and corresponding to a level of PDP configuration for each gear of the performance configuration; real-time counting the occupancy rate of performance resource currently used by the user in the performance configuration purchased by the user; judging whether the occupancy rate exceeds a set threshold, if yes, proceeding to the next step, if not, ending; reading the performance configuration currently purchased by the user, and obtaining the PDP configuration corresponding to the performance configuration purchased by the user according to the constraint relationship; judging whether the current PDP configuration of the user and the corresponding PDP configuration are the same, if yes, ending; if not, reducing a level of PDP configuration on the basis of the current PDP configuration or reducing the current PDP configuration to the corresponding PDP configuration, and configuring all CPUs of the user according to the reduced PDP configuration.

[0008] Preferably, the PDP configuration comprises a first PDP configuration related to CPU performance and / or a second PDP configuration related to memory performance, the performance configuration comprises CPU configuration and / or memory configuration; the constraint relationship between the performance configuration purchased by the user and the PDP configuration comprises a constraint relationship between the CPU core number purchased by the user and the first PDP configuration and / or a constraint relationship between the memory size purchased by the user and the second PDP configuration; the occupancy rate comprises a first occupancy rate of CPU power consumption currently used by the user in the CPU configuration purchased by the user and / or a second occupancy rate of memory occupancy currently used by the user in the memory configuration purchased by the user.

[0009] Specifically, setting the constraint relationship between the performance configuration purchased by the user and the PDP configuration comprises the steps of: dividing the CPU configuration into multiple gears according to the CPU core number purchased by the user, corresponding to a level of first PDP configuration for each gear of the CPU configuration, and the higher the CPU configuration, the better the performance of the first PDP configuration; and / or dividing the memory configuration into multiple gears according to the memory size purchased by the user, corresponding to a level of second PDP configuration for each gear of the memory configuration, and the higher the memory configuration, the better the performance of the second PDP configuration.

[0010] More preferably, the PDP configuration comprises a first PDP configuration related to CPU performance and a second PDP configuration related to memory performance, and the performance configuration comprises CPU configuration and memory configuration; when setting the constraint relationship between the performance configuration purchased by the user and the PDP configuration: taking the CPU configuration as a first coordinate and establishing a first coordinate axis x, taking the memory configuration as a second coordinate and establishing a second coordinate axis y, establishing a performance constraint coordinate system, and setting the PDP configuration corresponding to the performance configuration (x, y) of each gear of the user.

[0011] Preferably, the highest level performance configuration corresponds to a PDP configuration without performance limitation, and each decrease of the performance configuration corresponds to at least one performance constraint of the PDP configuration.

[0012] Specifically, the PDP configuration comprises a first PDP configuration related to CPU performance and / or a second PDP configuration related to memory performance, and the performance configuration comprises CPU configuration and / or memory configuration; the highest level CPU performance corresponds to the first PDP configuration without CPU performance limitation, and each decrease of the CPU performance corresponds to at least one performance constraint of the first PDP configuration; and / or, the highest level memory performance corresponds to the second PDP configuration without memory access bandwidth limitation, and each decrease of the memory performance corresponds to at least one performance constraint of the second PDP configuration.

[0013] More specifically, the first PDP configuration is implemented based on micro-architecture level, and the performance constraint of the first PDP configuration comprises disabling transmission queue, reducing transmission queue virtual large, and disabling speculation execution; and / or, the second PDP configuration is implemented based on micro-architecture level, and the performance constraint of the second PDP configuration comprises reducing instruction prefetching and reducing instruction out-of-order execution.

[0014] Preferably, the user is assigned with a preset initial PDP configuration each time the user logs in the arm server cloud; when the user initially logs in the arm server cloud, the user is assigned with the highest PDP configuration as the initial PDP configuration, and after all the CPUs assigned to the user are configured according to the decreased PDP configuration, the current decreased PDP configuration is recorded as the initial PDP configuration. Thus, when the user does not use the performance resources of the arm server cloud in a large amount, the user can enjoy the good experience brought by the high PDP configuration, and when the user uses the performance resources in a large amount, the user is limited from using the public resources in a large amount in order to balance the fairness and prevent the user from affecting the experience of the user with high configuration.

[0015] More preferably, when the system reaches a preset period or the performance resource occupancy rate of the arm server cloud is lower than a preset occupancy rate, the initial PDP configuration of each user is updated to the highest PDP configuration.

[0016] The application further provides an arm server cloud service resource balancing system, which comprises a processor, a memory, and one or more operation instructions stored in the memory and executed by the processor to implement the arm server cloud service resource balancing method.

[0017] Compared with the prior art, the performance configuration is divided into multiple gears, and then the corresponding PDP configuration is matched for different performance configurations. When the performance resource occupancy rate used by the user does not exceed the set threshold, the user has a good performance experience, and when the performance resource occupancy rate used by the user exceeds the set threshold, the PDP configuration of the user is reduced or the PDP configuration of the user reaches the corresponding minimum PDP configuration, so as to limit the resource usage rate of the user, effectively utilize the cloud resources, maintain the fairness of the cloud service, protect the user experience of the user with high configuration, and prevent the long-term high-power usage of the arm server cloud. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 is a structural diagram of a SOC system of an arm server cloud.

[0019] Figure 2 is a flowchart of a method for balancing server cloud service resources in embodiment 1 of the application.

[0020] Figure 3 is a flowchart of a method for balancing server cloud service resources in embodiment 2 of the application.

[0021] Figure 4 is a performance coordinate of the constraint relationship between CPU configuration, memory configuration and PDP configuration.

[0022] Figure 5 is a structural block diagram of a system for balancing server cloud service resources. DETAILED DESCRIPTION

[0023] To explain the technical content, structural features, achieved purposes and effects of the application in detail, the following will be described in detail in combination with the embodiments and the accompanying drawings.

[0024] Reference Figure 1 The SOC system of the arm server cloud includes a CPU processing module 101, a DDR storage module 102, an SCP control module 103, and a system bus 104, which communicates between the CPU processing module 101 and the DDR storage module 102 and the SCP control module 103.

[0025] For example, the CPU processing module 101 includes multiple CPUs, and the DDR storage module 102 provides 500G of memory.

[0026] Embodiment 1 Reference Figure 2 The application discloses a method for balancing server cloud service resources, including steps S1 to S6.

[0027] Before step S1, a constraint relationship between the performance configuration and the PDP configuration purchased by the user is set and stored in advance: the performance configuration is divided into multiple gears according to the size of the performance configuration purchased by the user, and the performance configuration of each gear corresponds to a level of PDP configuration.

[0028] PDP, Performance Defined Power, performance defined power. This feature limits CPU performance and the memory access bandwidth of the CPU from the micro-architecture perspective.

[0029] In this embodiment, the performance configuration includes CPU configuration, and the PDP configuration includes a first PDP configuration related to CPU performance. The constraint relationship between the performance configuration and the PDP configuration of the user includes the constraint relationship between the CPU core number purchased by the user and the first PDP configuration.

[0030] The constraint relationship between the performance configuration and the PDP configuration purchased by the user includes the constraint relationship between the CPU core number purchased by the user and the first PDP configuration.

[0031] For example, the CPU configuration is divided into a first gear corresponding to 2 cores, a second gear corresponding to 4 cores, a third gear corresponding to 8 cores, and a fourth gear corresponding to 16 cores. In this embodiment, the fourth gear is the highest gear of the CPU configuration.

[0032] Preferably, the PDP configuration corresponding to the highest gear performance configuration does not limit performance, and the PDP configuration corresponding to each lower gear performance configuration has at least one performance constraint.

[0033] Preferably, the first PDP configuration corresponding to the highest gear CPU performance does not limit CPU performance, and the first PDP configuration corresponding to each lower gear CPU performance has at least one performance constraint.

[0034] Specifically, the first PDP configuration is implemented based on the micro-architecture level and includes constraints on performance points such as disabling the transmit queue, reducing the transmit queue virtual large, and disabling the speculative execution.

[0035] S1, real-time statistics of the user's current use of performance resources in the performance configuration purchased by the user.

[0036] Specifically, the first PDP configuration is implemented based on the micro-architecture level and includes constraints on performance points such as disabling the transmit queue, reducing the transmit queue virtual large, and disabling the speculative execution.

[0037] For example, the user purchases an 8-core CPU, which can provide a total power consumption of X1, and the total power consumption of all the CPUs currently used by the user is X2, and the first occupancy rate is X2 / X1.

[0038] S2, determining whether the occupancy rate of the user exceeds a set threshold.

[0039] Specifically, it is determined whether the CPU power consumption currently used by the user exceeds the first set threshold corresponding to the CPU occupancy rate in the CPU configuration purchased by the user.

[0040] If yes, step S3 is performed, and if no, the process ends.

[0041] S3, reading the performance configuration currently purchased by the user. Specifically, the CPU configuration purchased by the user is read.

[0042] S4, obtaining the PDP configuration corresponding to the performance configuration of the user according to the constraint relationship. Specifically, the first PDP configuration corresponding to the CPU configuration purchased by the user is obtained according to the constraint relationship.

[0043] S5, determining whether the current PDP configuration of the user and the corresponding PDP configuration are the same. If yes, the process ends, which means that the current PDP configuration of the user has reached its own minimum PDP configuration. If no, step S6 is performed.

[0044] Specifically, it is determined whether the current first PDP configuration of the user and the corresponding first PDP configuration are the same. If yes, the process ends. If no, step S6 is performed.

[0045] S6, reducing the PDP configuration by one level based on the current PDP configuration or reducing the current PDP configuration to the corresponding PDP configuration.

[0046] Specifically, reducing the first PDP configuration by one level based on the current first PDP configuration or reducing the current first PDP configuration to the corresponding first PDP configuration.

[0047] Specifically, it is determined whether the current power consumption of the servers (servers sharing the same SLC resource are classified as servers in one group) in the group to which the user belongs in the current arm server cloud exceeds a preset threshold, and if yes, the current PDP configuration is reduced to the corresponding PDP configuration, and if no, the PDP configuration is reduced by one level based on the current PDP configuration. In this application, the corresponding PDP configuration is the PDP configuration corresponding to the purchased performance configuration.

[0048] For example, the user purchases a 2-core CPU configuration, and the corresponding PDP configuration is a first-gear CPU configuration. If the current CPU configuration is a third-gear CPU configuration, the CPU configuration is directly reduced to a second-gear CPU configuration. In another embodiment, the user purchases a 2-core CPU configuration, and the corresponding PDP configuration is a first-gear CPU configuration. If the current CPU configuration is a fourth-gear CPU configuration, the CPU configuration is directly reduced to a first-gear CPU configuration corresponding to a first PDP configuration. In yet another embodiment, the user purchases a 2-core CPU configuration, and the corresponding PDP configuration is a first-gear CPU configuration. If the current CPU configuration is a third-gear CPU configuration, and the current power consumption of the server of the group to which the user belongs exceeds a preset threshold, the CPU configuration is directly reduced to a first-gear CPU configuration corresponding to a first PDP configuration. If the current power consumption of the server of the group to which the user belongs does not exceed the preset threshold, the CPU configuration is directly reduced to a second-gear CPU configuration corresponding to a first PDP configuration.

[0049] Preferably, the user is assigned a preset initial PDP configuration each time the user logs in to the arm server cloud. When the user initially logs in to the arm server cloud, the user is assigned a highest PDP configuration as the initial PDP configuration. After all CPUs of the current user are configured according to the reduced PDP configuration, the current reduced PDP configuration is recorded as the initial PDP configuration.

[0050] More preferably, the initial PDP configuration of each user is updated to the highest PDP configuration when the system reaches a preset period or the performance resource occupancy rate of the arm server cloud is lower than a preset occupancy rate.

[0051] Embodiment 2 Reference Figure 3 The application discloses a method for balancing server cloud service resources.

[0052] Different from embodiment 1, in embodiment 2, the PDP configuration includes a second PDP configuration related to memory performance. The performance configuration purchased by the user includes a memory configuration. The constraint relationship between the performance configuration and the PDP configuration of the user includes a constraint relationship between the memory size purchased by the user and the second PDP configuration.

[0053] The constraint relationship between the performance configuration and the PDP configuration purchased by the user is set by the following steps: the memory configuration is divided into multiple gears according to the memory size purchased by the user. The memory configuration of each gear corresponds to a second PDP configuration of a level, and the higher the memory configuration, the better the performance of the second PDP configuration.

[0054] For example, the memory configuration (memory configuration) is divided into 4G corresponding to the first gear, 8G corresponding to the second gear, 16G corresponding to the third gear and 32G corresponding to the fourth gear. In this embodiment, the fourth gear is the highest gear of the memory configuration.

[0055] In step S1, the second occupancy rate of the memory occupancy currently used by the user in the memory configuration purchased by the user is counted in real time.

[0056] In step S2, it is determined whether the second occupancy rate exceeds the second set threshold corresponding to the storage occupancy rate. If yes, step S3 is executed, and if no, the process ends.

[0057] In step S3, the memory configuration purchased by the user is read.

[0058] In step S4, the second PDP configuration corresponding to the memory configuration of the user is obtained according to the constraint relationship.

[0059] In step S5, it is determined whether the current second PDP configuration of the user and the corresponding second PDP configuration are the same.

[0060] In step S6, the second PDP configuration of one level lower than the current second PDP configuration is obtained or the current second PDP configuration is reduced to the corresponding second PDP configuration.

[0061] Specifically, it is determined whether the proportion of the memory occupancy of the server (servers sharing the same DDR resource are classified as servers in one group) of the group to which the user belongs in the total DDR storage in the current arm server cloud exceeds a preset threshold, and if yes, the current second PDP configuration is reduced to the corresponding second PDP configuration, and if no, the second PDP configuration of one level lower than the current second PDP configuration is obtained.

[0062] For example, the user purchases an 8G memory configuration, and the corresponding PDP configuration is the second gear memory configuration. If the current memory configuration is the fourth gear memory configuration, the memory configuration is directly reduced to the third gear memory configuration corresponding to the second PDP configuration. In another embodiment, the user purchases an 8G memory configuration, and the corresponding PDP configuration is the second gear memory configuration. If the current memory configuration is the fourth gear memory configuration, the memory configuration is directly reduced to the second gear memory configuration corresponding to the second PDP configuration. In yet another embodiment, the user purchases an 8G memory configuration, and the corresponding PDP configuration is the second gear memory configuration corresponding to the second PDP configuration. If the current memory configuration is the fourth gear memory configuration, and the memory occupancy of the server of the group to which the user belongs in the total DDR storage exceeds a preset threshold, the memory configuration is directly reduced to the second gear memory configuration corresponding to the second PDP configuration. If the preset threshold is not exceeded, the memory configuration is directly reduced to the third gear memory configuration corresponding to the second PDP configuration.

[0063] Embodiment 3 Based on Embodiment 1 and Embodiment 2, in Embodiment 3, the constraint relationship between the performance configuration purchased by the user and the PDP configuration includes the following steps: the CPU configuration is divided into multiple gears according to the number of CPU cores purchased by the user, and each gear of the CPU configuration corresponds to a level of the first PDP configuration, and the higher the CPU configuration, the better the performance of the first PDP configuration; the memory configuration is divided into multiple gears according to the size of the memory purchased by the user, and each gear of the memory configuration corresponds to a level of the second PDP configuration, and the higher the memory configuration, the better the performance of the second PDP configuration.

[0064] Reference Figure 4 When setting the constraint relationship between the performance configuration purchased by the user and the PDP configuration, the CPU configuration is taken as the first coordinate and a first coordinate axis x is established, the memory configuration is taken as the second coordinate y and a second coordinate axis y is established, a performance constraint coordinate system is established, and the PDP configuration corresponding to the performance configuration (x, y) of each gear of the user is set. The PDP configuration includes the first PDP configuration and the second PDP configuration.

[0065] For example, referring to Figure 4 The fourth gear corresponds to coordinate 0, the third gear corresponds to coordinate 1, the second gear corresponds to coordinate 2, and the first gear corresponds to coordinate 3. The larger the gear, the smaller the PDP configuration limit, and the smaller the coordinate value. The smaller the gear, the larger the PDP configuration limit, and the larger the coordinate value.

[0066] If a user (referred to as user A) purchases a CPU of 16 cores and a DDR memory of 4G, the configuration coordinate of user A is (0, 3), and the PDP configuration is the fourth gear of 16 cores corresponding to the first PDP configuration, and the first gear of 3G corresponding to the second PDP configuration.

[0067] The above covers 16 application scenarios from two dimensions of XY axes, and can effectively alleviate the performance degradation problem caused by mixed deployment of cloud scenarios and business resource competition.

[0068] In embodiment 3, the PDP configuration includes a first PDP configuration related to CPU performance and a second PDP configuration related to memory performance. The performance configuration purchased by the user includes CPU configuration and memory configuration. The constraint relationship between the performance configuration of the user and the PDP configuration includes a constraint relationship between the number of CPU cores purchased by the user and the first PDP configuration, and a constraint relationship between the size of the memory purchased by the user and the second PDP configuration.

[0069] For example, the CPU configuration (number of processing core configurations) is divided into a first gear corresponding to 2 cores, a second gear corresponding to 4 cores, a third gear corresponding to 8 cores, and a fourth gear corresponding to 16 cores. In this embodiment, the fourth gear is the highest gear of the CPU configuration. The memory configuration (size of the memory purchased) is divided into a first gear corresponding to 4G, a second gear corresponding to 8G, a third gear corresponding to 16G, and a fourth gear corresponding to 32G. In this embodiment, the fourth gear is the highest gear of the memory configuration.

[0070] The application discloses a method for realizing balanced distribution of server cloud service resources, which comprises steps S1 to S62.

[0071] In S1, the first occupancy rate of the CPU power consumption currently used by the user in the CPU configuration purchased by the user is statistically obtained in real time, and the second occupancy rate of the memory occupancy currently used by the user in the memory configuration purchased by the user is statistically obtained in real time.

[0072] In S2, it is determined whether the first occupancy rate exceeds a first set threshold value, if yes, step S31 is executed, and if not, the process is ended; it is determined whether the second occupancy rate exceeds a second set threshold value, if yes, step S32 is executed, and if not, the process is ended.

[0073] In step S31, the CPU configuration purchased by the user is read, and step S41 is executed.

[0074] In step S32, the memory configuration purchased by the user is read, and step S42 is executed.

[0075] In S41, the first PDP configuration corresponding to the CPU configuration of the user is obtained according to the constraint relationship, and step S51 is executed.

[0076] In S42, the second PDP configuration corresponding to the memory configuration of the user is obtained according to the constraint relationship, and step S52 is executed.

[0077] S51, determining whether the current first PDP configuration of the user is same as the corresponding first PDP configuration, if yes, executing step S61, if not, ending.

[0078] S52, determining whether the current second PDP configuration of the user is same as the corresponding second PDP configuration, if yes, executing step S62, if not, ending.

[0079] S61, reducing the current first PDP configuration by one level or reducing the current first PDP configuration to the corresponding first PDP configuration.

[0080] S62, reducing the current second PDP configuration by one level or reducing the current second PDP configuration to the corresponding second PDP configuration.

[0081] Reference Figure 5 The application further discloses an implementation server cloud service resource balancing system, which comprises a processor 21, a memory 22, one or more operation instructions 23, the one or more operation instructions 23 are stored in the memory 22 and are executed by the processor 21 to implement the server cloud service resource balancing method.

[0082] DSU: DynamIQ Shared Unit, a new multi-core management system unit. SOC: System on a Chip, a system on a chip, which integrates multiple cores and on-chip peripherals. SCP: System Control Processor, a microprocessor for soc system control and power management. Cache: cache, access speed is much faster than accessing DDR. LLC: Last Level Cache, the last level cache, the cache closest to the DDR controller. SLC: System Level Cache, system-level cache, also known as L3 Cache, which is the same concept as LLC. RDT: Resource Director Technology, which provides LLC and memory bandwidth allocation and monitoring capabilities. MPAM: Memory System Resource Partitioning and Monitoring, a memory resource management solution on SOC. PDP: Performance Defined Power, a micro-architecture level CPU performance and memory bandwidth access restriction.

[0083] The above disclosure is merely the preferred embodiments of the present application and is not intended to limit the scope of the present application. Any equivalent changes made in accordance with the scope of the present application should be covered by the scope of the present application.

Claims

1. A method for achieving resource balancing in cloud-based server services, characterized in that: include: The constraints between the performance configuration purchased by the user and the PDP configuration are pre-set and stored: the performance configuration is divided into multiple levels according to the size of the performance configuration purchased by the user, and each level of performance configuration corresponds to a level of PDP configuration. Real-time statistics on the utilization rate of performance resources currently used by users within their purchased performance configuration; Determine whether the occupancy rate exceeds a set threshold. If yes, proceed to the next step; otherwise, end. Read the performance configuration currently purchased by the user, and obtain the PDP configuration corresponding to the performance configuration purchased by the user based on the constraint relationship; Determine whether the user's current PDP configuration is the same as the corresponding PDP configuration; if so, end the process. Otherwise, reduce the PDP configuration by one level from the current PDP configuration, or reduce the current PDP configuration to the corresponding PDP configuration, and configure all CPUs of the user according to the reduced PDP configuration.

2. The method for achieving server cloud service resource balancing as described in claim 1, characterized in that: The PDP configuration includes a first PDP configuration related to CPU performance and / or a second PDP configuration related to memory performance. The performance configuration includes CPU configuration and / or memory configuration. The constraint relationship between the performance configuration and the PDP configuration purchased by the user includes the constraint relationship between the number of CPU cores purchased by the user and the first PDP configuration and / or the constraint relationship between the size of memory purchased by the user and the second PDP configuration. The utilization rate includes a first utilization rate of the CPU power consumption currently used by the user in its purchased CPU configuration and / or a second utilization rate of the memory occupancy currently used by the user in its purchased memory configuration.

3. The method for achieving server cloud service resource balancing as described in claim 2, characterized in that: Setting the constraint relationship between the performance configuration purchased by the user and the PDP configuration includes the following steps: dividing the CPU configuration into multiple tiers according to the number of CPU cores purchased by the user, with each tier of CPU configuration corresponding to a first-level PDP configuration, and the higher the CPU configuration, the better the performance of the first PDP configuration; and / or dividing the memory configuration into multiple tiers according to the memory size purchased by the user, with each tier of memory configuration corresponding to a second-level PDP configuration, and the higher the memory configuration, the better the performance of the second PDP configuration.

4. The method for achieving server cloud service resource balancing as described in claim 3, characterized in that: The PDP configuration includes a first PDP configuration related to CPU performance and a second PDP configuration related to memory performance. The performance configuration includes CPU configuration and memory configuration. When setting the constraint relationship between the performance configuration purchased by the user and the PDP configuration: the CPU configuration is used as the first coordinate and a first coordinate axis x is established, the memory configuration is used as the second coordinate and a second coordinate axis y is established, a performance constraint coordinate system is established, and the PDP configuration corresponding to each level of the user's performance configuration (x, y) is set.

5. The method for achieving server cloud service resource balancing as described in claim 1, characterized in that: No performance limit is imposed in the PDP configuration corresponding to the highest performance level. For each performance level that is reduced, at least one performance constraint is added to the corresponding PDP configuration.

6. The method for achieving server cloud service resource balancing as described in claim 5, characterized in that: The PDP configuration includes a first PDP configuration related to CPU performance and / or a second PDP configuration related to memory performance, wherein the performance configuration includes CPU configuration and / or memory configuration. The first PDP configuration corresponding to the highest CPU performance level does not impose CPU performance limitations. For each CPU performance level that decreases by one tier, the corresponding first PDP configuration adds at least one performance constraint. And / or, the second PDP configuration corresponding to the highest memory performance level does not impose memory access bandwidth limitations. For each memory performance level that decreases by one tier, the corresponding second PDP configuration adds at least one performance constraint.

7. The method for achieving server cloud service resource balancing as described in claim 6, characterized in that: The first PDP configuration is implemented at the microarchitecture level, and its performance constraints include disabling the send queue, reducing the virtual size of the send queue, and disabling speculative execution; and / or, The second PDP configuration is implemented at the microarchitecture level, and the performance constraints of the second PDP configuration include reducing instruction prefetching and reducing out-of-order instruction execution.

8. The method for achieving server cloud service resource balancing as described in claim 1, characterized in that: Each time a user logs into the ARM server cloud, a preset initial PDP configuration is assigned to the user. When a user first logs into the ARM server cloud, the user is assigned the highest PDP configuration as the initial PDP configuration. All CPUs currently allocated to the user are configured according to the reduced PDP configuration, and the reduced PDP configuration is recorded as the initial PDP configuration.

9. The method for achieving business resource balancing as described in claim 8, characterized in that: When the system reaches a preset period or when the performance resource utilization rate of the ARM server cloud is lower than a preset utilization rate, the initial PDP configuration of each user will be updated to the highest PDP configuration.

10. A system for balancing server cloud business resources, characterized in that: The method includes a processor, a memory, and one or more operation instructions, wherein the one or more operation instructions are stored in the memory and executed by the processor as described in any one of claims 1-9 to implement the server cloud service resource balancing method.

Images