Intelligent computing power scheduling method and system based on dynamic planning

Abstract

The invention belongs to the technical field of data processing, and particularly relates to an intelligent computing power scheduling method and system based on dynamic programming, which assesses the balance of resource allocation and determines a preliminary computing power grouping scheme by monitoring current available resource information of a plurality of computing power clusters in real time and calculating resource difference values among the clusters. And on the basis, matching task types suitable for each group according to historical task types and resource consumption characteristics, and adjusting a preliminary grouping scheme to form an optimized computing power group. And calculating the service capability score of each group according to a preset service level protocol by using the optimized computing power group, and providing a recommended computing power resource service set for the user after sorting according to the scores. According to the method, computing power grouping is optimized through a dynamic evaluation mechanism, resource allocation balance is ensured, different types of task requirements can be met, and the overall utilization rate and service quality of computing power resources are improved.

Description

Technical Field

[0001] The present invention belongs to the field of data processing technology, and specifically relates to an intelligent computing power scheduling method and system based on dynamic programming. Background Art

[0002] In the current data center and cloud computing environment, the management and scheduling of computing resources are crucial to ensuring service quality and improving resource utilization. Existing computing scheduling methods mainly rely on preset rules or simple load balancing algorithms to allocate resources. These methods usually first collect the resource usage of each computing cluster, and then determine the allocation of tasks based on certain strategies (such as polling, minimum number of connections, etc.). However, when faced with a complex multi-cluster environment, such methods are prone to unbalanced resource allocation problems, especially when there are significant resource differences between clusters. Traditional scheduling strategies often cannot effectively balance the load of each cluster, resulting in some resources being idle while other resources are overloaded.

[0003] A key problem in existing technologies is how to effectively evaluate the differences and make reasonable resource allocation decisions when there are large resource differences between multiple computing clusters. If the resource differences between clusters cannot be accurately measured, it may not be possible to achieve optimal resource utilization during task scheduling, which in turn affects the performance and service level of the entire system. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent computing power scheduling method and system based on dynamic programming, which monitors the resource usage of the cluster in real time and calculates the resource difference value between clusters to evaluate the balance of resource allocation, so as to solve the problems raised in the above background technology.

[0005] To achieve the above objectives, the present invention proposes, on one hand, an intelligent computing power scheduling method based on dynamic programming, comprising:

[0006] S1: Determine the currently available resource information of multiple computing power clusters, and calculate the resource difference value between each cluster based on the currently available resource information; S2: Evaluate the balance of resource allocation between each cluster based on the resource difference value, and determine a preliminary computing power grouping scheme according to the balance of resource allocation; S3: On the basis of the preliminary grouping scheme, match the task type suitable for each group according to the historical task type and resource consumption characteristics; S4: Adjust the preliminary grouping scheme in combination with the matching results of the task type to form an optimized computing power grouping; S5: Utilize the optimized computing power grouping to calculate the service capability score of each group according to the preset service level agreement; S6: Sort the optimized computing power grouping according to the service capability score, and provide the user with the sorted computing power grouping as a computing power resource service set recommendation.

[0007] Preferably, determining currently available resource information of multiple computing power clusters includes:

[0008] Collect data on processor utilization, memory usage, and network bandwidth usage of nodes in each cluster;

[0009] Apply Formula R i =α·CPU i +β·MEM i +γ·NEF i Calculate the comprehensive resource score R of each cluster i i , where CPU i Indicates processor utilization, MEM i Indicates memory usage, NEF i represents the network bandwidth utilization, and the coefficients α, β, and γ represent the weights of the processor, memory, and network in the comprehensive evaluation respectively;

[0010] According to the comprehensive resource score R i , compare the resource utilization between different clusters to determine the resource difference value D ij =R i -R j |, where D ij Represents the resource difference between cluster i and cluster j.

[0011] Preferably, calculating the resource difference value between the clusters according to the currently available resource information includes:

[0012] CPU utilization for each cluster i , memory usage MEM i and network bandwidth utilization rate NEF i Perform standardization and obtain the standardized value sCPU i ,MEM i ,sNET i ;

[0013] Use the formula R i ′=α·sCPU i +β·sMEM i +γ·sNET i Update the comprehensive resource score R of each cluster i i ′, to reflect the importance of the resource after normalization;

[0014] Calculate the resource difference D between any two clusters i and j i ' j =R i ′-R j ′|, which is used as the basis for adjusting the initial computing power grouping plan.

[0015] Preferably, evaluating the balance of resource allocation among clusters based on the resource difference value includes:

[0016] The resource difference value D of all clusters i ' j Summarize and calculate the average resource variance value

[0017] Using the formula To quantify the overall balance E of resource allocation among clusters, where N is the total number of clusters;

[0018] Whether the resource allocation reaches a predetermined balance standard is determined based on the size of the overall balance degree E.

[0019] Preferably, a preliminary computing power grouping scheme is determined according to the balance of resource allocation, including:

[0020] When the overall balance degree E is lower than a predetermined threshold, the resource difference value D i ' j Close clusters are classified into the same group;

[0021] For each group, define the average resource score of the clusters in the group Ensure that each group Minimize the differences;

[0022] By formula To evaluate the consistency of resource distribution within each group g, select S g The minimized combination is used as the preliminary computing power grouping scheme.

[0023] Preferably, the task types suitable for each group are matched according to the historical task types and resource consumption characteristics, including:

[0024] Analyze historical task records and extract typical resource consumption patterns corresponding to task types. t ;

[0025] Compare the resource characteristics of each computing power group g with the typical resource consumption pattern P t , applying the similarity measure Sim(P t ,R g ) to evaluate the matching degree;

[0026] Select the similarity measure Sim(P t ,R g )The highest task type t is used as the task type recommended for processing by the group g.

[0027] Preferably, adjusting the preliminary grouping scheme in combination with the matching result of the task type to form an optimized computing power grouping includes:

[0028] For each group g, the resource matching score M is recalculated based on the resource requirements of its recommended task type t. g =Sim(P t ,R g );

[0029] If the resource matching score of a group g is M g If it is lower than the preset threshold, the clusters in the group are reallocated to other groups to improve the resource matching score;

[0030] After adjustments, the final optimized computing power grouping plan is formed, so that the clusters in each group can better support their recommended task types.

[0031] Preferably, the service capability score of each group is calculated according to a preset service level agreement, including:

[0032] According to the requirements of the preset service level agreement, define the service quality indicator Q for each optimized group g g , including but not limited to response time, throughput and reliability;

[0033] Apply Formula S g =w1·Q response,g +w2·Q throughput,g +w3·Q reliability,g , calculate the service capability score S of each group g g , where w1, w2, w3 are the weights of response time, throughput and reliability respectively;

[0034] Based on the service capability score S g , assessing each group’s ability to provide services.

[0035] Preferably, providing the user with the computing power grouping after the sorting as a computing power resource service set recommendation, including:

[0036] Construct a service capability score based on S g The group sorting function F(S) outputs the sorted group sequence;

[0037] Apply the sorting function F(S) to all optimized groups to obtain the sorted group sequence G sorted ;

[0038] Traverse the sorted group sequence G sorted , select the first n groups as the candidate recommendation set;

[0039] Aggregate the information of the candidate recommendation set into a recommendation list RecList, where RecList includes but is not limited to a group identifier, a service capability score, and a recommendation reason;

[0040] The recommendation list RecList is displayed to the user so that the user can select a suitable computing resource service set according to his / her needs.

[0041] On the other hand, the present invention proposes an intelligent computing power scheduling system based on dynamic programming, comprising:

[0042] A cluster resource status monitoring module is used to determine the currently available resource information of multiple computing power clusters and calculate the resource difference value between the clusters based on the currently available resource information;

[0043] A resource balance assessment and preliminary grouping module, used to assess the balance of resource allocation among clusters based on the resource difference value, and determine a preliminary computing power grouping plan according to the balance of resource allocation;

[0044] A task type matching module is used to match the task type suitable for each group according to the historical task type and resource consumption characteristics based on the preliminary grouping scheme;

[0045] A grouping optimization adjustment module, used to adjust the preliminary grouping scheme in combination with the matching result of the task type to form an optimized computing power grouping;

[0046] A service capability score calculation module, used to calculate the service capability score of each group according to a preset service level agreement by using the optimized computing power grouping;

[0047] The computing power resource service recommendation module is used to sort the optimized computing power groups according to the service capability scores, and provide the users with the sorted computing power groups as a computing power resource service set recommendation.

[0048] Technical effects and advantages of the present invention: Compared with the prior art, the intelligent computing power scheduling method and system based on dynamic programming proposed by the present invention has the following advantages:

[0049] The present invention evaluates the balance of resource allocation by monitoring the resource usage of the cluster in real time and calculating the resource difference value between clusters. Based on this dynamic evaluation mechanism, the computing power grouping scheme can be determined more reasonably, and through further task type matching and grouping optimization, it is ensured that the computing power grouping finally formed can better meet the needs of different types of tasks. In addition, by introducing a service level agreement to calculate the service capability score of each group and making ranking recommendations based on it, the overall utilization rate and service quality of computing power resources can be effectively improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a flow chart of the intelligent computing power scheduling method based on dynamic programming of the present invention;

[0051] Figure 2 The block diagram of the intelligent computing power scheduling system based on dynamic programming of the present invention. DETAILED DESCRIPTION

[0052] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. The specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0053] The present invention provides an intelligent computing power scheduling method based on dynamic programming, which aims to evaluate the balance of resource allocation and determine a preliminary computing power grouping plan by monitoring the current available resource information of multiple computing power clusters in real time and calculating the resource difference values ​​between clusters.

[0054] On this basis, the task types suitable for each group are matched according to the historical task types and resource consumption characteristics, and the preliminary grouping scheme is adjusted to form an optimized computing power grouping. Using the optimized computing power grouping, the service capability score of each group is calculated according to the preset service level agreement, and the recommended computing power resource service set is provided to the user after sorting by the score. The present invention optimizes the computing power grouping through a dynamic evaluation mechanism to ensure that resources are allocated evenly and can meet the needs of different types of tasks, thereby improving the overall utilization and service quality of computing power resources. The details are as follows:

[0055] like Figure 1 As shown, in this embodiment, the intelligent computing power scheduling method based on dynamic programming includes the following steps:

[0056] S1: Determine the currently available resource information of multiple computing power clusters, and calculate the resource difference values ​​between each cluster based on the currently available resource information; by determining the currently available resource information of multiple computing power clusters in real time, and calculating the resource difference values ​​between each cluster accordingly, the present invention can accurately reflect the resource usage and differences between each cluster. This resource difference value calculation method based on real-time data helps to more accurately evaluate the actual load conditions of different clusters, thereby providing a solid foundation for subsequent resource allocation and scheduling. In this way, waste and excessive concentration of resources can be avoided, ensuring that computing power resources are used more evenly and effectively, thereby improving the operating efficiency and service quality of the overall system.

[0057] In addition, determining the currently available resource information of multiple computing power clusters further includes:

[0058] Collect the processor utilization, memory utilization and network bandwidth utilization data of the nodes in each cluster; specifically, regularly collect the processor utilization, memory utilization and network bandwidth utilization data from each node in each computing cluster. For example, a node in cluster A may report that its processor utilization is 80%, memory utilization is 60%, and network bandwidth utilization is 40%.

[0059] Apply Formula R i =α·CPU i +β·MEM i +γ·NEF i Calculate the comprehensive resource score R of each cluster i i , where CPU i Indicates processor utilization, MEM i Indicates memory usage, NEF i represents the network bandwidth utilization, and the coefficients α, β, and γ represent the weights of the processor, memory, and network in the comprehensive evaluation respectively;

[0060] According to the comprehensive resource score R i , compare the resource utilization between different clusters to determine the resource difference value D ij =|R i -R j |, where D ij Represents the resource difference between cluster i and cluster j.

[0061] For example: Assuming α=0.5, β=0.3, γ=0.2, for cluster A, if its average processor utilization CPUA is 80%, memory utilization MEMA is 60%, and network bandwidth occupancy NETA is 40%, then its comprehensive resource score RA is 0.5×80+0.3×60+0.2×40=64.

[0062] Through the above steps, the present invention can accurately obtain the current resource usage of each computing power cluster, and obtain a comprehensive resource score through weighted calculation, which reflects the overall resource utilization level of the cluster. Furthermore, by comparing the comprehensive resource scores of different clusters, the resource differences between clusters can be clarified, which provides a reliable basis for subsequent resource allocation. This method not only improves the accuracy of resource evaluation, but also simplifies the quantification process of resource differences, making resource scheduling more refined and automated, thereby improving the overall computing power resource utilization efficiency and service quality.

[0063] Calculating the resource difference value between clusters according to the currently available resource information further includes:

[0064] CPU utilization for each cluster i , memory usage MEMi and network bandwidth utilization rate NEF i Perform standardization and obtain the standardized value sCPU i ,MEM i ,sNET i ;

[0065] Use the formula R i ′=α·sCPU i +β·sMEM i +γ·sNET i Update the comprehensive resource score R of each cluster i i ′, to reflect the importance of standardized resources; calculate the resource difference value D between any two clusters i and j i ' j =R i ′-R j ′|, which is used as the basis for adjusting the initial computing power grouping plan.

[0066] By standardizing the resource utilization data of each cluster, the present invention eliminates the dimensional differences between different resource types, so that resource utilization can be compared under a unified standard. The updated comprehensive resource score more accurately reflects the importance of the current resource usage of each cluster, which helps to more accurately evaluate the resource differences between clusters. The calculated resource difference value provides basic data support for subsequent computing power grouping optimization, ensuring that the grouping scheme is more reasonable, thereby improving the balance of resource allocation and the overall service level of the system.

[0067] S2: Evaluate the balance of resource allocation among clusters based on the resource difference value, and determine a preliminary computing power grouping scheme according to the balance of resource allocation; By evaluating the balance of resource allocation among computing power clusters based on the resource difference value, and determining a preliminary computing power grouping scheme accordingly, the present invention can ensure that resource usage within each group is more uniform. This grouping method helps to reduce the risk of overload of certain clusters, while making full use of those underutilized cluster resources. Therefore, this method can improve the balance and utilization of the entire system resources, reduce service bottlenecks caused by improper resource allocation, and thus improve service stability and user experience.

[0068] Specifically, evaluating the balance of resource allocation among clusters based on the resource difference value further includes:

[0069] The resource difference value D of all clusters i ' j Summarize and calculate the average resource variance value For example:

[0070] Assume there are four clusters A, B, C, and D, and their resource difference values ​​are D AB =0.04, D AC =0.12, D AD =0.08, D BC =0.10, D BD =0.06, D CD =0.05.

[0071] Sum up all resource variance values ​​and calculate the average resource variance value:

[0072]

[0073] Using the formula To quantify the overall balance E of resource allocation among clusters, where N is the total number of clusters;

[0074] For example: Assume D i ' j represents the standardized resource difference between cluster i and cluster j. Here, D is used directly. ij .

[0075] The overall balance degree E = 1 / 4(4-1)(∣0.04-0.0767∣+∣0.12-0.0767∣+∣0.08-0.0767∣+∣0.10-0.0767∣+∣0.06-0.0767∣+∣0.05-0.0767∣).

[0076] E=1 / 12(0.0367+0.0433+0.0033+0.0233+0.0167+0.0267)=0.0228.

[0077] Whether the resource allocation has reached the predetermined balance standard is determined according to the size of the overall balance degree E. Assuming that the predetermined balance standard threshold is 0.05, since E=0.0228<0.05, the resource allocation has reached the predetermined balance standard.

[0078] Through the above steps, the present invention can effectively evaluate the balance of resource allocation between clusters and determine the preliminary computing power grouping plan accordingly. The advantage of doing so is that it can ensure a more even resource allocation, reduce resource waste, and improve the efficiency of computing power resource utilization. By dividing clusters with similar resource difference values ​​into the same group and ensuring that the resource distribution within the group is consistent, the utilization of computing power resources can be further optimized, and the overall performance and service quality of the system can be improved.

[0079] Determining a preliminary computing power grouping scheme according to the balance of resource allocation further includes:

[0080] When the overall balance degree E is lower than a predetermined threshold, the resource difference value D i ' j Close clusters are grouped together; as shown above, due to D AD = 0.08D and D CD = 0.05 is small, so we can consider grouping clusters A and D, and clusters C and D into one group. Assume that A and D are selected as one group, C and D are selected as another group, and the remaining B is a separate group. For each group, define the average resource score of the clusters in the group Ensure that each group The difference between To evaluate the consistency of resource distribution within each group g, select S g The minimized combination is used as the preliminary computing power grouping scheme.

[0081] Through the above steps, the present invention can determine a preliminary computing power grouping scheme based on the balance of resource allocation. This method ensures that clusters with similar resource difference values ​​are divided into the same group, thereby improving the consistency of resource distribution within the group, and further improving the balance and efficiency of resource use. g As a preliminary computing power grouping solution, the combination can better meet the needs of different task types and improve the performance and service quality of the entire system.

[0082] S3: Based on the preliminary grouping scheme, the task types suitable for each group are matched according to the historical task types and resource consumption characteristics; the details are as follows:

[0083] Analyze historical task records and extract typical resource consumption patterns corresponding to task types. t ; Compare the resource characteristics of each computing power group g with the typical resource consumption pattern P t , applying the similarity measure Sim(P t ,R g ) to evaluate the matching degree; select the similarity measure Sim(P t ,R g )The highest task type t is used as the task type recommended for processing by the group g.

[0084] By matching the task types suitable for each group based on the historical task types and resource consumption characteristics on the basis of the preliminary grouping scheme, the present invention can achieve efficient matching between tasks and computing resources. This method can identify the best resource allocation for different types of tasks based on past task execution, so that when allocating new tasks, they can be assigned to the computing group that is most suitable for processing such tasks.

[0085] Doing so will not only improve the efficiency and success rate of task processing, but also further optimize resource utilization, reduce resource waste, and improve the performance and service quality of the overall system.

[0086] S4: adjusting the preliminary grouping scheme in combination with the matching result of the task type to form an optimized computing power grouping; specifically, as follows: including:

[0087] For each group g, the resource matching score Mg=Sim(Pt,Rg) is recalculated based on the resource requirements of its recommended task type t; if the resource matching score M of a group g is g If it is lower than the preset threshold, the clusters in the group are reallocated to other groups to improve the resource matching score. After adjustment, the final optimized computing power grouping plan is formed, so that the clusters in each group can better support their recommended task types.

[0088] By adjusting the preliminary grouping scheme in combination with the matching results of the task types to form optimized computing power groups, the present invention can ensure that the computing power resources in each group are more consistent with the task types to be processed. This method not only improves the pertinence and efficiency of task execution, but also makes resource allocation more reasonable, reducing performance bottlenecks and resource waste caused by resource mismatch.

[0089] The optimized computing power grouping can better adapt to the needs of different task types, thereby improving the overall response speed and service quality of the system, while ensuring the maximum efficiency of resource utilization.

[0090] S5: Using the optimized computing power grouping, calculate the service capability score of each group according to the preset service level agreement; specifically as follows: According to the requirements of the preset service level agreement, define the service quality indicator Q for each optimized group g g , including but not limited to response time, throughput and reliability; applying formula S g =w1·Q response,g +w2·Q throughput,g +w3·Q reliability,g , calculate the service capability score S of each group g g , where w1, w2, w3 are the weights of response time, throughput and reliability respectively; based on the service capability score S g , assessing each group’s ability to provide services.

[0091] By using the optimized computing power grouping and calculating the service capability score of each grouping according to the preset service level agreement, the present invention can quantify the service capabilities of different computing power groups. This method allows managers to objectively evaluate the performance of each group in terms of response time, throughput, reliability, etc., so as to better understand the current service level of the system.

[0092] Based on the service capability score, computing resources can be managed and scheduled more scientifically, ensuring that important tasks are prioritized while improving user satisfaction and overall service quality. In addition, the scoring mechanism can also help identify potential service shortcomings and provide a basis for further optimizing the system.

[0093] S6: Sort the optimized computing power groups according to the service capability scores, and provide the users with the sorted computing power groups as a computing power resource service set recommendation. The details are as follows:

[0094] Construct a service capability score based on S g The group sorting function F(S) is used to output the sorted group sequence. The sorting function F(S) is applied to all optimized groups to obtain the sorted group sequence G. sorted ; Traverse the sorted group sequence G sorted , select the first n groups as candidate recommendation sets; summarize the information of the candidate recommendation sets into a recommendation list RecList, where RecList includes but is not limited to group identifiers, service capability scores and recommendation reasons; display the recommendation list RecList to the user so that the user can select a suitable computing resource service set according to his own needs.

[0095] By sorting the optimized computing power groups according to the service capability scores and providing the sorted computing power groups to users as computing power resource service set recommendations, the present invention realizes intelligent management of computing power resources.

[0096] This method allows users to intuitively understand the service capabilities and availability of different computing power groups, making it easier for users to choose the most appropriate computing power resources according to their needs. At the same time, the recommendation mechanism based on score sorting ensures that high-quality resources are recommended first, thereby improving the user experience when selecting and using computing power services, and enhancing service reliability and user satisfaction.

[0097] In addition, this approach also helps promote the efficient use of resources and reduce the waste of resources caused by blind selection.

[0098] On the other hand, the present invention proposes an intelligent computing power scheduling system based on dynamic programming, such as Figure 2 As shown, it includes: cluster resource status monitoring module, resource balance assessment and preliminary grouping module, task type matching module, grouping optimization and adjustment module, service capability scoring calculation module and computing power resource service recommendation module.

[0099] Specifically, the cluster resource status monitoring module is used to determine the currently available resource information of multiple computing power clusters, and calculate the resource difference value between the clusters according to the currently available resource information;

[0100] Specifically, the resource balance evaluation and preliminary grouping module is used to evaluate the balance of resource allocation between clusters based on the resource difference value, and determine a preliminary computing power grouping plan according to the balance of resource allocation;

[0101] Specifically, the task type matching module is used to match the task type suitable for each group according to the historical task type and resource consumption characteristics based on the preliminary grouping scheme;

[0102] Specifically, the grouping optimization adjustment module is used to adjust the preliminary grouping scheme in combination with the matching result of the task type to form an optimized computing power grouping;

[0103] Specifically, the service capability score calculation module is used to calculate the service capability score of each group according to a preset service level agreement by using the optimized computing power grouping;

[0104] Specifically, the computing power resource service recommendation module is used to sort the optimized computing power groups according to the service capability scores, and provide the sorted computing power groups to the user as a computing power resource service set recommendation.

[0105] In addition, the above-mentioned cluster resource status monitoring module, resource balance assessment and preliminary grouping module, task type matching module, grouping optimization adjustment module, service capability score calculation module and computing power resource service recommendation module are also used to implement the other steps of the above-mentioned dynamic programming-based intelligent computing power scheduling method during execution, which will not be elaborated here one by one.

[0106] In addition, the present invention also provides a terminal device. The intelligent computing power scheduling method based on dynamic programming involved in this embodiment is mainly applied to the terminal device, which can be a PC, a portable computer, a mobile terminal, or other device with display and processing functions.

[0107] Specifically, the terminal device may include a processor (such as a CPU), a communication bus, a user interface, a network interface, and a memory. The communication bus is used to realize the connection and communication between these components; the user interface may include a display screen (Displ ay), an input unit such as a keyboard (Keyboard); the network interface may optionally include a standard wired interface, a wireless interface (such as a WIFI interface); the memory may be a high-speed RAM memory, or a stable memory (non-volatile memory), such as a disk memory, and the memory may optionally be a storage device independent of the aforementioned processor.

[0108] Among them, the memory stores a readable storage medium, and the readable storage medium stores a computing power scheduling program. The processor can call the computing power scheduling program stored in the memory and execute the intelligent computing power scheduling method based on dynamic programming provided by an embodiment of the present invention.

[0109] It is understood that a readable storage medium can be a tangible device that can hold and store instructions used by an instruction execution device. Computer-readable storage media can be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the above. More specific examples of computer-readable storage media (a non-exhaustive list) include: portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), static random access memories (SRAM), portable compact disk read-only memories (CD-ROM), digital versatile disks (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or raised structures in grooves on which instructions are stored, and any suitable combination of the above. The computer-readable storage medium used here is not to be interpreted as a transient signal itself, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through a wire.

[0110] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0111] The computer program instructions for performing the operations of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer, partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect via the Internet). In some embodiments, by utilizing state information of computer-readable program instructions to personalize an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), the electronic circuit can execute the computer-readable program instructions to implement various aspects of the present disclosure.

[0112] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent computing power scheduling method based on dynamic programming, characterized in that: include: S1: Determine currently available resource information of multiple computing power clusters, and calculate resource difference values ​​between the clusters based on the currently available resource information; S2: Evaluate the balance of resource allocation among clusters based on the resource difference value, and determine a preliminary computing power grouping plan according to the balance of resource allocation; S3: Based on the preliminary grouping scheme, matching the appropriate task type for each group according to the historical task type and resource consumption characteristics; S4: adjusting the preliminary grouping scheme in combination with the matching result of the task type to form an optimized computing power grouping; S5: Using the optimized computing power grouping, calculate the service capability score of each group according to a preset service level agreement; S6: Sort the optimized computing power groups according to the service capability scores, and provide the sorted computing power groups to the user as a computing power resource service set recommendation.

2. According to claim 1, a method for intelligent computing power scheduling based on dynamic programming is characterized in that: Determine the currently available resource information of multiple computing clusters, including: Collect data on processor utilization, memory usage, and network bandwidth usage of nodes in each cluster; Apply Formula R i =α·CPU i +β·MEM i +γ·NEF i Calculate the comprehensive resource score R of each cluster i i , where CPU i Indicates processor utilization, MEM i Indicates memory usage, NEF i represents the network bandwidth utilization, and the coefficients α, β, and γ represent the weights of the processor, memory, and network in the comprehensive evaluation respectively; According to the comprehensive resource score R i , compare the resource utilization between different clusters to determine the resource difference value D ij =|R i -R j |, where D ij Represents the resource difference between cluster i and cluster j.

3. According to claim 2, the intelligent computing power scheduling method based on dynamic programming is characterized in that: Calculating resource difference values ​​between clusters according to the currently available resource information includes: CPU utilization for each cluster i , memory usage MEM i and network bandwidth utilization rate NEF i Perform standardization and obtain the standardized value sCPU i ,MEM i ,sNET i ; Use the formula R′ i =α·sCPU i +β·sMEM i +γ·sNET i Update the comprehensive resource score R′ of each cluster i i , to reflect the importance of the resource after normalization; Calculate the resource difference D′ between any two clusters i and j ij =|R′ i -R′ j |, and use this as the basis for adjusting the initial computing power grouping plan.

4. According to claim 3, the intelligent computing power scheduling method based on dynamic programming is characterized in that: Evaluating the balance of resource allocation among clusters based on the resource difference value includes: The resource difference value D of all clusters i ' j Summarize and calculate the average resource variance value Using the formula To quantify the overall balance E of resource allocation among clusters, where N is the total number of clusters; Whether the resource allocation reaches a predetermined balance standard is determined based on the size of the overall balance degree E.

5. According to claim 4, the intelligent computing power scheduling method based on dynamic programming is characterized in that: According to the balance of resource allocation, a preliminary computing power grouping scheme is determined, including: When the overall balance degree E is lower than a predetermined threshold, the resource difference value D i ' j Close clusters are classified into the same group; For each group, define the average resource score of the clusters in the group Ensure that each group Minimize the differences; By formula To evaluate the consistency of resource distribution within each group g, select S g The minimized combination is used as the preliminary computing power grouping scheme.

6. The intelligent computing power scheduling method based on dynamic programming according to claim 5 is characterized in that: Match the appropriate task type for each group based on historical task types and resource consumption characteristics, including: Analyze historical task records and extract typical resource consumption patterns corresponding to task types. t ; Compare the resource characteristics of each computing power group g with the typical resource consumption pattern P t , applying the similarity measure Sim(P t ,R g ) to evaluate the matching degree; Select the similarity measure Sim(P t ,R g )The highest task type t is used as the task type recommended for processing by the group g.

7. The intelligent computing power scheduling method based on dynamic programming according to claim 6 is characterized in that: The preliminary grouping scheme is adjusted in combination with the matching result of the task type to form an optimized computing power grouping, including: For each group g, the resource matching score M is recalculated based on the resource requirements of its recommended task type t. g =Sim(P t ,R g ); If the resource matching score of a group g is M g If it is lower than the preset threshold, the clusters in the group are reallocated to other groups to improve the resource matching score; After adjustments, the final optimized computing power grouping plan is formed, so that the clusters in each group can better support their recommended task types.

8. The intelligent computing power scheduling method based on dynamic programming according to claim 7 is characterized in that: Calculate the service capability score of each group based on the preset service level agreement, including: According to the requirements of the preset service level agreement, define the service quality indicator Q for each optimized group g g , including but not limited to response time, throughput and reliability; Apply Formula S g =w1·Q response,g +w2·Q throughput,g +w3·Q reliability,g , calculate the service capability score S of each group g g , where w1, w2, w3 are the weights of response time, throughput and reliability respectively; Based on the service capability score S g , assessing each group’s ability to provide services.

9. The intelligent computing power scheduling method based on dynamic programming according to claim 1 is characterized in that: Providing the user with the computing power groupings sorted as a computing power resource service set recommendation, including: Construct a service capability score based on S g The group sorting function F(S) outputs the sorted group sequence; Apply the sorting function F(S) to all optimized groups to obtain the sorted group sequence G sorted ; Traverse the sorted group sequence G sorted , select the first n groups as the candidate recommendation set; Aggregate the information of the candidate recommendation set into a recommendation list RecList, where RecList includes but is not limited to a group identifier, a service capability score, and a recommendation reason; The recommendation list RecList is displayed to the user so that the user can select a suitable computing resource service set according to his / her needs.

10. An intelligent computing power scheduling system based on dynamic programming, characterized in that it includes: A cluster resource status monitoring module is used to determine the currently available resource information of multiple computing power clusters and calculate the resource difference value between the clusters based on the currently available resource information; A resource balance assessment and preliminary grouping module, used to assess the balance of resource allocation among clusters based on the resource difference value, and determine a preliminary computing power grouping plan according to the balance of resource allocation; A task type matching module is used to match the task type suitable for each group according to the historical task type and resource consumption characteristics based on the preliminary grouping scheme; A grouping optimization adjustment module, used to adjust the preliminary grouping scheme in combination with the matching result of the task type to form an optimized computing power grouping; A service capability score calculation module, used to calculate the service capability score of each group according to a preset service level agreement by using the optimized computing power grouping; The computing power resource service recommendation module is used to sort the optimized computing power groups according to the service capability scores, and provide the users with the sorted computing power groups as a computing power resource service set recommendation.

Images