Computing power node scheduling optimization method and device, equipment, medium and product

Abstract

This invention discloses a method, apparatus, device, medium, and product for optimizing computing power node scheduling. The method includes quantifying the allocable computing power of each node in a computing power cluster; obtaining a set of nodes that meet the computing power threshold requirements based on the quantified cluster computing power matrix and a preset computing power threshold matrix; and using a greedy binary search algorithm based on the node set and target business requirements to find a subset of nodes corresponding to each type of computing power that meets the target business requirements, thereby optimizing computing power node scheduling. This method can quickly filter the remaining computing power matrix using the computing power quantification method and calculate each node that meets each type of computing power requirement based on the greedy binary search algorithm, reducing the computational cost and algorithm complexity of computing power node scheduling.

Description

Technical Field

[0001] This invention relates to the field of computing power network scheduling technology, and in particular to a method, apparatus, equipment, medium and product for optimizing computing power node scheduling. Background Technology

[0002] In computing power networks, the allocation of computing power for applications such as cloud gaming, cloud rendering, cloud AR / VR, and cloud phones typically requires existing methods for optimizing computing node scheduling. These methods usually involve selecting suitable computing nodes to run computational tasks based on actual computing power demands. For example, user business needs are broken down into specific computing power requirements. Resources are matched among idle nodes in the computing power resource pool. Node computing power, storage, and bandwidth resources are identified as a matrix, and matching nodes are selected. Then, a scheduling model is used to calculate path latency to determine the shortest latency path, or a publish-subscribe communication model is established with edge devices through a distributed database. The target node and forwarding path are determined by combining the node's computing power status and link communication status. While existing methods can meet computing power scheduling needs in most cases, they are computationally expensive and have high algorithm complexity. Summary of the Invention

[0003] This invention provides a method, apparatus, device, medium, and product for optimizing computing power node scheduling. It quickly filters the remaining computing power matrix through a computing power quantification method and calculates each node that meets the computing power requirements for each type based on a greedy binary search algorithm, thereby reducing the computational cost and algorithm complexity of computing power node scheduling.

[0004] To achieve the above objectives, embodiments of the present invention provide a method for optimizing computing node scheduling, comprising: The allocable computing power of each node in the computing power cluster is quantified, and the set of nodes that meet the computing power threshold requirements is obtained based on the quantified cluster computing power matrix and the preset computing power threshold matrix. Based on the node set and the target business requirements, a greedy binary search algorithm is used to find the node subset corresponding to each type of computing power that meets the target business requirements, so as to optimize the computing power node scheduling.

[0005] As an improvement to the above scheme, the step of quantifying the allocable computing power of each node in the computing power cluster, and obtaining the set of nodes that meet the computing power threshold requirements based on the quantized cluster computing power matrix and the preset computing power threshold matrix, includes: The allocable computing power of each node in the computing power cluster is quantized by performing a first-order matrix mapping, resulting in the quantized cluster computing power matrix. The set of nodes that meet the computing power threshold requirements is obtained based on the quantized cluster computing power matrix and the preset computing power threshold matrix.

[0006] As an improvement to the above scheme, the step of obtaining the set of nodes that meet the computing power threshold requirements based on the quantized cluster computing power matrix and the preset computing power threshold matrix includes: The quantized cluster computing power matrix and the preset computing power threshold matrix are compared according to each type of computing power. Nodes whose quantized computing power values ​​for all types of computing power in the quantized cluster computing power matrix are greater than or equal to the corresponding computing power threshold in the computing power threshold matrix are considered as nodes that meet the computing power threshold requirements, thus obtaining a set of nodes that meet the computing power threshold requirements.

[0007] As an improvement to the above scheme, the step of using a greedy binary search algorithm to find the subset of nodes corresponding to each type of computing power that satisfies the target business requirements based on the node set and target business needs, in order to optimize computing power node scheduling, includes: Based on the node set and the target business requirements, a greedy binary search algorithm is used to find the node subset corresponding to each type of computing power that satisfies the target business requirements. The optimal computing power node is determined by using an intersection weighted algorithm for each subset of nodes, in order to optimize the scheduling of computing power nodes.

[0008] As an improvement to the above scheme, the step of using a greedy binary search algorithm to find the subset of nodes corresponding to each type of computing power that satisfies the target business requirements based on the node set and the target business requirements includes: The node set is sorted in ascending order according to each type of computing power to obtain a node subset sorted by each type of computing power. Based on the target business requirements, a greedy binary search algorithm is used to search for the node subsets after sorting each type of computing power, so as to obtain the node subsets corresponding to each type of computing power that meet the target business requirements.

[0009] As an improvement to the above scheme, the step of determining the optimal computing power node based on the intersection weighted algorithm for each node subset to optimize computing power node scheduling includes: Find the intersection of each subset of nodes to obtain a set of candidate nodes corresponding to all types of computing power that meet the target business requirements; The computing power of each node in the candidate node set is weighted according to a preset computing power weighting value to obtain the optimal computing power node, so as to optimize the scheduling of computing power nodes.

[0010] To achieve the above objectives, embodiments of the present invention provide a computing node scheduling optimization device, comprising: The computing power node quantization module is used to quantify the allocable computing power of each node in the computing power cluster, and obtain the set of nodes that meet the computing power threshold requirements based on the quantized cluster computing power matrix and the preset computing power threshold matrix. The computing power node search module is used to search using a greedy binary search algorithm based on the node set and target business requirements to obtain a subset of nodes corresponding to each type of computing power that meets the target business requirements, so as to optimize the computing power node scheduling.

[0011] To achieve the above objectives, this invention provides a computing node scheduling optimization device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the above-mentioned computing node scheduling optimization method.

[0012] To achieve the above objectives, embodiments of the present invention also provide a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the above-described computing node scheduling optimization method.

[0013] To achieve the above objectives, embodiments of the present invention also provide a computer program product, which is stored in a storage medium and executed by at least one processor to implement the steps of the above-described computing node scheduling optimization method.

[0014] Compared with existing technologies, the present invention discloses a computing power node scheduling optimization method, apparatus, device, medium, and product. This method quantifies the allocable computing power of each node in a computing power cluster. Based on the quantified cluster computing power matrix and a preset computing power threshold matrix, a set of nodes meeting the computing power threshold requirements is obtained. A greedy binary search algorithm is then used to search for the nodes that meet each type of computing power requirement within the target business needs, thereby optimizing the computing power node scheduling. By quantifying the computing power of the cluster and combining it with the computing power threshold matrix, the set of nodes meeting the basic computing power requirements is quickly filtered out, directly eliminating nodes that do not meet the threshold, significantly reducing the number of nodes participating in subsequent algorithm calculations. Furthermore, the greedy binary search algorithm used to find the subset of computing power nodes based on the filtered node set reduces the computational cost and algorithm complexity of computing power node scheduling. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating a computing node scheduling optimization method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a cloud mobile phone service structure provided by an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of a computing node scheduling optimization device provided in an embodiment of the present invention; Figure 4 This is a structural block diagram of a computing node scheduling optimization device provided in an embodiment of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] It should be noted that the terms "comprising" and "specific" in this invention, and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or device.

[0018] Please see Figure 1 , Figure 1 This is a flowchart illustrating a computing node scheduling optimization method provided in an embodiment of the present invention. The computing node scheduling optimization method includes: S1, quantify the allocable computing power of each node in the computing power cluster, and obtain the set of nodes that meet the computing power threshold requirements based on the quantized cluster computing power matrix and the preset computing power threshold matrix. S2, based on the node set and the target business requirements, a greedy binary search algorithm is used to search for the node subset corresponding to each type of computing power that meets the target business requirements, so as to optimize the computing power node scheduling.

[0019] For example, the computing power node scheduling optimization method described in this embodiment of the invention can be implemented by a computing power scheduling server, which is capable of information interaction with the target user. The computing power scheduling server obtains the allocatable computing power of each node in the computing power cluster, maps the allocatable computing power of each node to a first-order matrix, obtaining a quantized cluster computing power matrix. The computing power of each node in the quantized cluster computing power matrix is ​​compared with a preset computing power threshold matrix for each type of computing power. If the quantized value of all computing power types of a node is greater than or equal to the corresponding computing power threshold, then the node meets the computing power threshold requirement; otherwise, it is discarded, resulting in a set of nodes that meet the computing power threshold requirement. The actual business computing power requirement (target business requirement) is decomposed into a business requirement matrix. Based on the business requirement matrix, a greedy binary search algorithm is used to find a subset of nodes in the node set that meet each type of computing power in the target business requirement, in order to optimize the computing power node scheduling. This invention addresses the issue by mapping the cluster computing power matrix to a first-order matrix through computing power quantification. Combined with a computing power threshold matrix, it quickly filters out the set of nodes that meet the basic computing power requirements and directly eliminates nodes that do not meet the threshold, significantly reducing the number of nodes participating in subsequent algorithm calculations and reducing the amount of computation from the source. Then, based on the filtered set of nodes, a greedy binary search algorithm is used to find a subset of computing power nodes, reducing the algorithmic complexity of computing power scheduling and reducing additional computing power consumption.

[0020] Specifically, step S1 includes: S11, perform first-order matrix mapping on the allocable computing power of each node in the computing power cluster to quantize the computing power, and obtain the quantized cluster computing power matrix. S12, based on the quantized cluster computing power matrix and the preset computing power threshold matrix, obtain the set of nodes that meet the computing power threshold requirements.

[0021] For example, define a first-order matrix of computing power for a single node. ,in, Each corresponds to a specific type of computing power quantification value in a certain computing power set node; Indicates the first The computational power value; This represents the number of computing power types. The remaining computing power of the computing power cluster is mapped to a quantized cluster computing power matrix. A computing power threshold matrix is ​​defined based on the minimum computing power requirements of the target business for cluster nodes. To define the node computing power threshold, by using the remaining computing power matrix The computing power of each node With computing power threshold matrix The set of nodes that meet the computing power threshold requirement is obtained by comparing the computing power of each type. Among them, the quantized cluster computing power matrix... for: , Computing power threshold matrix for: , In the formula, Indicates the first The computing power of each node This represents the total number of nodes in the computing cluster. Indicates the first The node in the node The computational power value; Indicates the first A certain computing power threshold.

[0022] More specifically, step S12 includes: S121, compare the quantized cluster computing power matrix and the preset computing power threshold matrix according to each type of computing power; S122, nodes whose quantized values ​​of all types of computing power in the quantized cluster computing power matrix are greater than or equal to the corresponding type of computing power threshold in the computing power threshold matrix are taken as nodes that meet the computing power threshold requirements, so as to obtain a set of nodes that meet the computing power threshold requirements.

[0023] For example, by using the remaining computing power matrix (the quantized cluster computing power matrix). The computing power of each node With computing power threshold matrix The computing power of each type is compared. If the quantized value of a node's computing power for all types is greater than or equal to the corresponding computing power threshold, then the node meets the computing power threshold requirement; otherwise, it is removed. This yields a set of nodes that meet the computing power threshold requirement. : , , In the formula, Indicates the first The computing power of each node; Indicates the first The computing power verification results of each node, when When, it indicates the first The computing power of each node When the computing power threshold requirement is met, When, it indicates the first The computing power of each node The computing power threshold requirement is not met. Indicates the first The node in the node The computational power value, ; Indicates the first A computing power threshold. When , indicating the first The node in the node If the quantified computing power value meets the corresponding computing power threshold requirement, then the next computing power type is recursively calculated. The result is then ANDed with the previous value to obtain a value of 1, indicating that the node meets the computing power threshold requirement. If the computing power threshold... Greater than the The computing power of each node Corresponding computing power types When, it indicates that the node The computing power threshold requirement is not met. This invention can filter out nodes that meet the computing power threshold requirement, thereby reducing the number of nodes participating in the computation. This is particularly effective for clusters with a large number of computing power nodes, significantly reducing computation time.

[0024] Specifically, step S2 includes: S21, based on the node set and the target business requirements, a greedy binary search algorithm is used to search for the node subset corresponding to each type of computing power that satisfies the target business requirements. S22, the optimal computing power node is determined by the intersection weighting algorithm based on each node subset, so as to optimize the computing power node scheduling.

[0025] For example, the actual business computing power requirements (target business requirements) are broken down into a business requirement matrix. , ,in, Indicates the first Various computing power business needs, According to the business requirements matrix A greedy binary search algorithm is used on the set of nodes. Identify the subset of nodes corresponding to each type of computing power that meets the target business requirements. For each node subset, use an intersection-weighted algorithm to determine the optimal computing power node, and then optimize the computing power node scheduling based on the optimal computing power node.

[0026] More specifically, step S21 includes: S211, Sort the node set in ascending order according to the computing power of each type to obtain a node subset after sorting by computing power of each type; S212, according to the target business requirements, the greedy binary search algorithm is used to search the node subsets after sorting each type of computing power to obtain the node subsets that satisfy each type of computing power in the target business requirements.

[0027] For example, a set of nodes Each node is sorted in ascending order according to its computing power, resulting in... A sorted subset of nodes is denoted by the sorting function, and all ascending sorted results constitute the sorted set. For each sorted subset of nodes A greedy binary search algorithm is used to quickly find the first condition that satisfies the condition. Nodes with computing power service requirements (judgment criteria:) , Indicates the first (This refers to a computing power service requirement). This node and all subsequent nodes satisfy this computing power service requirement. Let this be denoted by a binary search function, and we obtain the result that satisfies the first... A subset of nodes required for various computing power services The subset of nodes that meet the computing power service requirements constitutes the search set. It is understandable that the greedy binary search algorithm deploys computing power requirements to the fewest nodes that can meet those requirements. From the perspective of the cluster as a whole, this maximizes the utilization of computing power nodes and supports the deployment of computing power requirements for more users. The time complexity of the greedy binary search algorithm is O(nlogn), which is significantly lower than the O(n^2) time complexity of the traditional recursive algorithm.

[0028] Among them, sorting function for: , Sort sets for: , Binary search function for: , Search set for: , In the formula, Indicates according to the first A subset of nodes sorted in ascending order of computing power. ; Indicates according to the first A subset of nodes sorted in ascending order of computing power; Indicates that the first A subset of nodes required for various computing power services; Indicates that the first A subset of nodes that meet the computing power service requirements.

[0029] More specifically, step S22 includes: S221, find the intersection of each node subset to obtain the candidate node set corresponding to all types of computing power that meet the target business requirements; S222, the computing power of each node in the candidate node set is weighted according to the preset computing power weighting value to obtain the optimal computing power node, so as to optimize the computing power node scheduling.

[0030] For example, since the search results for different types of computing power may contain the same nodes, and a single type of computing power may also have multiple sorted results, a subset of nodes that meet the computing power service requirements for each type can be used. Find the intersection to obtain the set of candidate nodes that meet the computing power service requirements of all types. :

[0031] Since different services have varying requirements for different types of computing power, the final scheduling priority can be set using a weighted method based on the characteristics of the target service. This means that the weighting values ​​are preset for each service type, and a weighted calculation function is defined. For the candidate node set For each node in the array, calculate its... The weighted sum of computing power by type serves as the overall score for that node: ,in, For the first The overall score of each node; Indicates the first The weighted value of computing power From the candidate node set The node with the highest overall score is selected as the optimal computing power node. .

[0032] No. The expression for the comprehensive score of each node is: , Optimal computing power node for: , In the formula, For the first The overall score of each node; Indicates the first The weighted value of computing power ; This indicates the optimal computing node that meets the computing power service requirements.

[0033] In a specific implementation, taking cloud phone services as an example, such as Figure 2 As shown, Figure 2This is a schematic diagram of a cloud phone service structure provided by an embodiment of the present invention; a cloud phone refers to a virtual phone deployed on a cloud server. Users can remotely control the cloud phone deployed on the cloud server through terminals such as App (Application) or Web (World Wide Web). The cloud phone relies on public cloud and ARM (Advanced RISC Machines) virtualization technology to provide users with an Android virtual phone in the cloud. Users can establish a connection via WebRTC and remotely control the cloud phone in real time via video streaming. The latency of the cloud phone is crucial to the user experience. The latency from when the user issues a command to when the terminal receives the video stream from the cloud phone mainly includes four parts: command acquisition latency, GPU (Graphics Processing Unit) video acquisition, encoding and rendering latency, encoding and transmission bitrate latency, and terminal decoding and rendering latency. Therefore, the computing power scheduling strategy will affect the encoding and transmission bitrate latency, and the shortest latency path algorithm will minimize the latency of this link to the greatest extent.

[0034] On the cloud server side, based on Android (Android or mobile operating system) container virtualization technology, AOSP (Android Open Source Project) images are loaded into Kubernetes (K8S, container orchestration system) pods. The virtualized physical hardware in a pod mainly includes an ARM-based CPU (Central Processing Unit), memory, GPU, and storage volume size. Therefore, the main factors affecting the service quality for cloud phone users are latency, CPU, memory, GPU, and storage volume. For cloud phone service providers, reducing latency helps improve the usability of cloud phones and enhance user experience. Increasing the utilization rate of physical hardware can further reduce infrastructure construction and control costs.

[0035] The allocable computing power matrix of the cluster nodes is mapped to first order, and no further division is made according to the type of computing power. ,in, These correspond to the CPU, GPU, storage volume, memory, and latency of a certain computing power node, respectively. Indicates the first The computational power value, when hour, Indicates the delay quantization value; This represents the number of computing power types. Assuming the computing power cluster has 5 nodes, the quantized values ​​of the computing power are as follows: , , , , Target business needs Computing power threshold matrix The set of nodes that meet the computing power threshold requirement is obtained by comparing each node with the computing power threshold matrix. . Some nodes are removed because their computing power does not meet the computing power threshold requirements.

[0036] The results are obtained by sorting according to each type of computing power: , , , , , Because the computing power is the same. There are two possible outcomes. Then, based on the binary search algorithm, the first node that satisfies the computing power business requirements for each type of computing power is found, resulting in a subset of nodes that satisfy each type of computing power business requirement: , , , , , in, middle and Further exclusion was made. The above results represent subsets of nodes that meet the respective computing power requirements of their respective services. The greedy strategy here is to deploy the user's computing power requirements onto the smallest number of nodes that can satisfy that requirement. For the entire cluster, the greedy strategy aims to deploy as many user computing power requirements as possible. The search results show that the same nodes exist after searching for different types of computing power, and even for a single computing power type, there are multiple sorted results. Therefore, the intersection method is used to obtain the set of candidate nodes that meet the requirements of all types of computing power: , in, and All of these methods can meet the user's computing power requirements. Theoretically, deployment on any node can satisfy the greedy strategy mentioned in step two. However, different services have different requirements for different types of computing power. For example, cloud phones have more stringent latency requirements. Therefore, a weighted method is used to set the final scheduling priority.

[0037] , , Finally, the optimal computing power node is determined through weighted calculation. The computing power network scheduling strategy of this invention optimizes the deployment of computing power to the optimal nodes while meeting user latency experience requirements. Unlike traditional scheduling strategies, it considers the cost of computing power service providers and improves the off-grid utilization of physical infrastructure. The advantage of this scheduling strategy is its low algorithm time complexity.

[0038] This invention discloses a method for optimizing computing power node scheduling. It quantifies the allocable computing power of each node in a computing power cluster, and obtains a set of nodes that meet the computing power threshold requirements based on the quantified cluster computing power matrix and a preset computing power threshold matrix. Then, a greedy binary search algorithm is used to search for the nodes that meet each type of computing power requirement within the target business needs, thereby optimizing computing power node scheduling. By quantifying the computing power, the cluster computing power is quickly filtered out using the computing power threshold matrix to identify the set of nodes that meet the basic computing power requirements, directly eliminating nodes that do not meet the threshold, significantly reducing the number of nodes participating in subsequent algorithm calculations. Furthermore, the greedy binary search algorithm used to find the subset of computing power nodes based on the filtered node set reduces the computational cost and algorithm complexity of computing power node scheduling.

[0039] See Figure 3 , Figure 3 This is a schematic diagram of a computing node scheduling optimization device 10 provided in an embodiment of the present invention. The computing node scheduling optimization device 10 includes: The computing power node quantization module 11 is used to quantify the allocable computing power of each node in the computing power cluster, and obtain the set of nodes that meet the computing power threshold requirements based on the quantized cluster computing power matrix and the preset computing power threshold matrix. The computing power node search module 12 is used to search using a greedy binary search algorithm based on the node set and the target business requirements to obtain a subset of nodes corresponding to each type of computing power that meets the target business requirements, so as to optimize the computing power node scheduling.

[0040] The computing node scheduling optimization device 10 provided in this embodiment of the invention can realize all the processes of the computing node scheduling optimization method of the above embodiment. The functions and technical effects of each module in the device are the same as the functions and technical effects of the computing node scheduling optimization method of the above embodiment, and will not be repeated here.

[0041] See Figure 4 , Figure 4This is a schematic diagram of the structure of a computing node scheduling optimization device 20 provided in an embodiment of the present invention. The computing node scheduling optimization device 20 of this embodiment includes: a processor 21, a memory 22, and a computer program stored in the memory 22 and executable on the processor 21. When the processor 21 executes the computer program, it implements the steps in the above-described computing node scheduling optimization method embodiment. Alternatively, when the processor 21 executes the computer program, it implements the functions of each module in the above-described computing node scheduling optimization device embodiment.

[0042] For example, the computer program may be divided into one or more modules, which are stored in the memory 22 and executed by the processor 21 to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the computing node scheduling and optimization device 20.

[0043] The computing node scheduling and optimization device 20 can be a desktop computer, laptop, handheld computer, or cloud server, etc. The computing node scheduling and optimization device 20 may include, but is not limited to, a processor 21 and a memory 22. Those skilled in the art will understand that the schematic diagram is merely an example of the computing node scheduling and optimization device 20 and does not constitute a limitation on the device. It may include more or fewer components than shown, or combine certain components, or use different components. For example, the computing node scheduling and optimization device 20 may also include input / output devices, network access devices, buses, etc.

[0044] The processor 21 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor. The processor 21 is the control center of the computing node scheduling and optimization device 20, connecting all parts of the computing node scheduling and optimization device 20 via various interfaces and lines.

[0045] The memory 22 can be used to store the computer programs and / or modules. The processor 21 implements various functions of the computing node scheduling and optimization device 20 by running or executing the computer programs and / or modules stored in the memory 22 and calling the data stored in the memory 22. The memory 22 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory 22 may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0046] If the modules integrated in the computing node scheduling optimization device 20 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by the processor 21, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content contained in the computer-readable medium may be appropriately added to or subtracted from the content as required by the legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium may not include electrical carrier signals and telecommunication signals.

[0047] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0048] This invention also provides a computer-readable storage medium, which includes a stored computer program, wherein the computer program, when running, controls the device where the computer-readable storage medium is located to execute the computing node scheduling optimization method as described in the above embodiments.

[0049] Furthermore, embodiments of the present invention also provide a computer program product, which is stored in a storage medium and executed by at least one processor to implement the steps of the computing node scheduling optimization method of the above embodiments.

[0050] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for optimizing the scheduling of computing nodes, characterized in that, include: The allocable computing power of each node in the computing power cluster is quantified, and the set of nodes that meet the computing power threshold requirements is obtained based on the quantified cluster computing power matrix and the preset computing power threshold matrix. Based on the node set and the target business requirements, a greedy binary search algorithm is used to find the node subset corresponding to each type of computing power that meets the target business requirements, so as to optimize the computing power node scheduling.

2. The computing node scheduling optimization method as described in claim 1, characterized in that, The process of quantifying the allocable computing power of each node in the computing power cluster, and obtaining the set of nodes that meet the computing power threshold requirements based on the quantized cluster computing power matrix and the preset computing power threshold matrix, includes: The allocable computing power of each node in the computing power cluster is quantized by performing a first-order matrix mapping, resulting in the quantized cluster computing power matrix. The set of nodes that meet the computing power threshold requirements is obtained based on the quantized cluster computing power matrix and the preset computing power threshold matrix.

3. The computing node scheduling optimization method as described in claim 2, characterized in that, The step of obtaining the set of nodes that meet the computing power threshold requirements based on the quantized cluster computing power matrix and the preset computing power threshold matrix includes: The quantized cluster computing power matrix and the preset computing power threshold matrix are compared according to each type of computing power. Nodes whose quantized computing power values ​​for all types of computing power in the quantized cluster computing power matrix are greater than or equal to the corresponding computing power threshold in the computing power threshold matrix are considered as nodes that meet the computing power threshold requirements, thus obtaining a set of nodes that meet the computing power threshold requirements.

4. The computing node scheduling optimization method as described in claim 1, characterized in that, The step of using a greedy binary search algorithm based on the node set and target business requirements to find a subset of nodes corresponding to each type of computing power that meets the target business requirements, in order to optimize computing power node scheduling, includes: Based on the node set and the target business requirements, a greedy binary search algorithm is used to find the node subset corresponding to each type of computing power that satisfies the target business requirements. The optimal computing power node is determined by using an intersection weighted algorithm for each subset of nodes, in order to optimize the scheduling of computing power nodes.

5. The computing node scheduling optimization method as described in claim 4, characterized in that, The step of using a greedy binary search algorithm based on the node set and target business requirements to find the node subset corresponding to each type of computing power that satisfies the target business requirements includes: The node set is sorted in ascending order according to each type of computing power to obtain a node subset sorted by each type of computing power. Based on the target business requirements, a greedy binary search algorithm is used to search for the node subsets after sorting each type of computing power, so as to obtain the node subsets corresponding to each type of computing power that meet the target business requirements.

6. The computing node scheduling optimization method as described in claim 4, characterized in that, The step of determining the optimal computing power node based on the intersection weighted algorithm for each subset of nodes, in order to optimize computing power node scheduling, includes: Find the intersection of each subset of nodes to obtain a set of candidate nodes corresponding to all types of computing power that meet the target business requirements; The computing power of each node in the candidate node set is weighted according to a preset computing power weighting value to obtain the optimal computing power node, so as to optimize the scheduling of computing power nodes.

7. A computing node scheduling optimization device, characterized in that, include: The computing power node quantization module is used to quantify the allocable computing power of each node in the computing power cluster, and obtain the set of nodes that meet the computing power threshold requirements based on the quantized cluster computing power matrix and the preset computing power threshold matrix. The computing power node search module is used to search using a greedy binary search algorithm based on the node set and target business requirements to obtain a subset of nodes corresponding to each type of computing power that meets the target business requirements, so as to optimize the computing power node scheduling.

8. A computing node scheduling optimization device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the computing node scheduling optimization method as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the computing node scheduling optimization method as described in any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product is stored in a storage medium, and the program product is executed by at least one processor to implement the steps of the computing node scheduling optimization method as described in any one of claims 1-6.

Images