Real-time task automatic scheduling management method and system

Abstract

The invention relates to the technical field of computers, and discloses a real-time task automatic scheduling management method and system.The method comprises the steps that a current task is obtained, subtasks of the current task are extracted, and a task operation feature set is determined according to the subtasks of the current task; clustering the sub-tasks according to the task operation feature set, and dividing the sub-tasks into corresponding clusters; obtaining a resource characteristic curve of the current resource node and a task characteristic curve of the subtasks in each cluster, and determining a target scheduling node of each cluster according to the resource characteristic curve and the task characteristic curve; and calculating scheduling parameters of each sub-task to the corresponding target scheduling node, and scheduling each sub-task according to the scheduling parameters. According to the method, the resource nodes of the tasks can be reasonably allocated in real time and efficiently, and the resource utilization rate is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, more particularly, to a real-time task automatic scheduling management method and system. BACKGROUND

[0002] Resource scheduling refers to the process of allocating and managing computing resources effectively in a computing environment according to predetermined strategies and algorithms. These computing resources include but are not limited to processor resources, memory, storage space, network bandwidth, etc. The goal of resource scheduling is to optimize resource utilization, improve system performance, reduce operating costs, and ensure that various tasks and services meet the established quality of service requirements.

[0003] The resource scheduling of the prior art is only limited to the scheduling of idle nodes. However, due to the different resource characteristics of each resource node, the task and the resource node may not be well adapted, and it is difficult to maximize the resource utilization of each resource node. SUMMARY

[0004] The present application provides a real-time task automatic scheduling management method and system to solve the problem that the task scheduling in the prior art cannot maximize the resource utilization of each resource node, comprising: Obtain the current task and extract the subtasks of the current task, determine the task running feature set according to the subtasks of the current task, cluster each subtask according to the task running feature set, divide each subtask into the corresponding cluster, obtain the resource feature curve of the current resource node and the task feature curve of the subtasks in each cluster, determine the target scheduling node of each cluster according to the resource feature curve and the task feature curve, calculate the scheduling parameters of each subtask to the corresponding target scheduling node, and schedule each subtask according to the scheduling parameters.

[0005] Further, the determination of the task running feature set according to the subtasks of the current task comprises: obtaining historical task running data, extracting the subtasks of each task according to the historical task running data, counting the resource occupation change of the subtasks, drawing the resource occupation curve according to the resource occupation change of the subtasks, determining the task running feature of the subtasks according to the resource occupation curve, establishing a training sample set according to the task running feature of the subtasks, establishing a running feature extraction model according to the training sample set and training the running feature extraction model to obtain a trained running feature extraction model, and inputting the subtasks of the current task into the trained running feature extraction model to obtain the task running feature set.

[0006] Further, the task running feature of the subtask is determined according to the resource occupation curve, comprising: obtaining a preset segmentation section, segmenting the resource occupation curve of the subtask according to the preset segmentation section to obtain a plurality of resource occupation curve segments; obtaining the maximum value and the minimum value of each resource occupation curve segment, calculating the difference between the maximum value and the minimum value of the resource occupation curve segment to obtain a first task running feature; obtaining the average value of each resource occupation curve segment to obtain a second task running feature; counting the number of resource occupation curve segments whose average value is greater than a preset standard average value to obtain a third task running feature; and determining the task running feature of the subtask according to the first task running feature, the second task running feature and the third task running feature.

[0007] Further, the task running feature of the subtask is determined according to the resource occupation curve, comprising: obtaining a preset segmentation section, segmenting the resource occupation curve of the subtask according to the preset segmentation section to obtain a plurality of resource occupation curve segments; obtaining the maximum value and the minimum value of each resource occupation curve segment, calculating the difference between the maximum value and the minimum value of the resource occupation curve segment to obtain a first task running feature; obtaining the average value of each resource occupation curve segment to obtain a second task running feature; counting the number of resource occupation curve segments whose average value is greater than a preset standard average value to obtain a third task running feature; and determining the task running feature of the subtask according to the first task running feature, the second task running feature and the third task running feature.

[0008] Further, the task running feature of the subtask is determined according to the resource occupation curve, comprising: obtaining a preset segmentation section, segmenting the resource occupation curve of the subtask according to the preset segmentation section to obtain a plurality of resource occupation curve segments; obtaining the maximum value and the minimum value of each resource occupation curve segment, calculating the difference between the maximum value and the minimum value of the resource occupation curve segment to obtain a first task running feature; obtaining the average value of each resource occupation curve segment to obtain a second task running feature; counting the number of resource occupation curve segments whose average value is greater than a preset standard average value to obtain a third task running feature; and determining the task running feature of the subtask according to the first task running feature, the second task running feature and the third task running feature.

[0009] wherein, is the feature distance, is the first task running feature of the i-th curve segment of the subtask, is the first task running feature of the i-th curve segment of the cluster center, is the second task running feature of the i-th curve segment of the subtask, is the second task running feature of the i-th curve segment of the cluster center, is the third task running feature of the subtask, is the third task running feature of the cluster center, is the number of curve segments, , , are respectively the first conversion coefficient, the second conversion coefficient and the third conversion coefficient, .

[0010] Further, the target scheduling node of each cluster is determined according to the resource characteristic curve and the task characteristic curve, comprising: calculating the correlation coefficient of the resource characteristic curve and the task characteristic curve, and determining the adaptation degree of each resource node to each cluster according to the correlation coefficient of the resource characteristic curve and the task characteristic curve; obtaining the adaptation degree of any cluster to each resource node, and sorting each resource node in descending order of the adaptation degree; screening the resource node satisfying the preset allowable order position according to the sorting result, obtaining the target scheduling node of the cluster, and counting the target scheduling nodes of all clusters.

[0011] Further, the scheduling parameter of each subtask to the corresponding target scheduling node is calculated, comprising: determining the maximum resource occupation of each subtask according to the resource occupation curve of each subtask, obtaining the real-time residual resource of the target scheduling node of the cluster to which the subtask belongs, and calculating the difference between the real-time resource availability and the maximum resource occupation; determining the resource availability of the target scheduling node according to the difference between the real-time resource availability and the maximum resource occupation, obtaining the adaptation degree of the cluster to which the subtask belongs to the target scheduling node, and determining the scheduling parameter of the subtask to the target scheduling node according to the adaptation degree and the resource availability, and counting the scheduling parameter of all subtasks to the corresponding target scheduling node.

[0012] Further, the scheduling parameter of the subtask to the target scheduling node is determined according to the adaptation degree and the resource availability, comprising: determining the scheduling parameter according to the scheduling parameter calculation formula, wherein the scheduling parameter is,

[0013] wherein, is the scheduling parameter, is the adaptation degree, is the resource availability, , respectively, is a preset first weight and a preset second weight.

[0014] Further, the subtasks are scheduled according to the scheduling parameter, comprising: obtaining a preset scheduling threshold, calculating the scheduling parameter of the subtask and the corresponding target scheduling node according to the sorting position; determining whether the scheduling parameter is greater than the preset scheduling threshold, if the scheduling parameter is greater than the preset scheduling threshold, the subtask is scheduled to the corresponding target scheduling node; if the scheduling parameter is less than or equal to the preset scheduling threshold, the scheduling parameter of the target scheduling node and the subtask of the next sorting position is continued to be calculated until all subtasks are completed.

[0015] In order to achieve the above purpose, the application further provides a real-time task automatic scheduling management system, comprising: The extraction module is used to acquire a current task and extract subtasks of the current task, determine a task running feature set according to the subtasks of the current task; the clustering module is used to cluster the subtasks according to the task running feature set, and divide the subtasks into corresponding clustering clusters; the target module is used to acquire a resource feature curve of a current resource node and a task feature curve of the subtasks in each clustering cluster, determine a target scheduling node of each clustering cluster according to the resource feature curve and the task feature curve; and the scheduling module is used to calculate scheduling parameters of each subtask to the corresponding target scheduling node, and schedule each subtask according to the scheduling parameters.

[0016] The application has the advantages that By applying the above technical solution, the current task is disassembled into multiple subtasks, and the subtasks with similar resource demand features are clustered, and each is matched to a node with the most suitable resource supply feature, thereby improving the adaptability of the resource node to the subtasks, enabling the resource node of the task to be reasonably allocated in real time and efficiently, and improving resource utilization. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0018] Figure 1 A general flowchart of a real-time task automatic scheduling management method according to an embodiment of the present application is shown; Figure 2 A structural schematic diagram of a real-time task automatic scheduling management system according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0020] An embodiment of the present application provides a real-time task automatic scheduling management method, as shown in Figure 1 The method comprises the following steps: S101, acquiring a current task and extracting subtasks of the current task, and determining a task running feature set according to the subtasks of the current task; In some embodiments of the present application, the task running feature set is determined according to the subtasks of the current task, including: obtaining historical task running data, extracting the subtasks of each task according to the historical task running data; counting the resource occupation change of the subtasks, drawing a resource occupation curve according to the resource occupation change of the subtasks; determining the task running feature of the subtasks according to the resource occupation curve, establishing a training sample set according to the task running feature of the subtasks; establishing a running feature extraction model according to the training sample set and training the running feature extraction model to obtain a trained running feature extraction model; inputting the subtasks of the current task into the trained running feature extraction model to obtain the task running feature set.

[0021] In the present embodiment, the historical task running data is extracted, the historical tasks are decomposed into subtasks of multiple types, and the resource occupation change of the subtasks in the execution process is counted. The resource occupation is specifically the CPU resource occupation. The resource occupation curve is drawn and the task running feature is calculated through the resource occupation change.

[0022] In the present embodiment, the historical subtasks and the corresponding task running features are preprocessed. The preprocessing at least includes one of the following: data cleaning, normalization, feature construction and feature extraction, to obtain a training sample set. The historical subtasks in the training sample set are taken as input variables, and the task running features are taken as output variables. A deep learning neural network model is established to obtain a running feature extraction model. The network structure of the running feature extraction model is ResNet50. 80% of the training sample set is taken as a training set, and 20% is taken as a test set. A mean square error loss function is used, and an optimizer uses Adam. After training, a trained running feature extraction model is finally obtained. The current task to be executed is obtained in real time and decomposed into subtasks of multiple types. The subtasks are input into the trained running feature extraction model to obtain the task running feature set of the subtasks.

[0023] In some embodiments of the present application, the task running feature of the subtask is determined according to the resource occupation curve, including: obtaining a preset segmentation section, segmenting the resource occupation curve of the subtask according to the preset segmentation section to obtain a plurality of resource occupation curve segments; obtaining the maximum value and the minimum value of each resource occupation curve segment, calculating the difference between the maximum value and the minimum value of the resource occupation curve segment to obtain a first task running feature; obtaining the average value of each resource occupation curve segment to obtain a second task running feature; counting the number of resource occupation curve segments whose average value is greater than a preset standard average value to obtain a third task running feature; and determining the task running feature of the subtask according to the first task running feature, the second task running feature and the third task running feature.

[0024] In this embodiment, the peak fluctuation characteristic of resource occupation is obtained by the difference between the maximum value and the minimum value of the resource occupation curve segment, the average occupation characteristic is obtained by the average value of the resource occupation curve segment, the occupation fluctuation characteristic is obtained by the number of resource occupation curve segments whose average value is greater than the preset standard average value, and the task running characteristic of the subtask is composed of the above three characteristics.

[0025] In S102, each subtask is clustered according to the task running characteristic set, and each subtask is divided into a corresponding cluster. In some embodiments of the present application, the clustering of each subtask according to the task running characteristic set comprises: establishing a sample data set according to each subtask of the current task, randomly selecting k initial cluster centers of the sample data set; calculating the feature distance of the subtask in the sample data set to the initial cluster center, and dividing each subtask into a corresponding cluster according to the feature distance of the subtask in the sample data set to the initial cluster center; calculating the average value of the task running characteristic of the subtask in each cluster, and reselecting the cluster center according to the average value of the feature distance of the subtask in each cluster; repeating the above steps until the cluster center no longer changes or the number of iterations reaches the preset maximum number of iterations, and obtaining the clustering result of the subtask.

[0026] In this embodiment, the k-means clustering algorithm is used to cluster the subtasks according to the feature distance between the subtasks, and the value of k of the algorithm is determined by the number of subtasks. In this embodiment, the value of k is set to 4.

[0027] In some embodiments of the present application, the calculation of the feature distance of the subtask in the sample data set to the initial cluster center comprises: determining the feature distance according to the feature distance calculation formula, wherein the feature distance calculation formula is,

[0028] wherein, is the feature distance, is the first task running characteristic of the i-th curve segment of the subtask, is the first task running characteristic of the i-th curve segment of the cluster center, is the second task running characteristic of the i-th curve segment of the subtask, is the second task running characteristic of the i-th curve segment of the cluster center, is the third task running characteristic of the subtask, is the third task running characteristic of the cluster center, is the number of curve segments, , , are respectively the first conversion coefficient, the second conversion coefficient and the third conversion coefficient, .

[0029] S103, acquire resource characteristic curves of the current resource nodes and task characteristic curves of the sub-tasks in each cluster, and determine target scheduling nodes of each cluster according to the resource characteristic curves and the task characteristic curves; In some embodiments of the present application, the determining of the target scheduling nodes of each cluster according to the resource characteristic curves and the task characteristic curves comprises: calculating correlation coefficients of the resource characteristic curves and the task characteristic curves, determining adaptation degrees of each resource node to each cluster according to the correlation coefficients of the resource characteristic curves and the task characteristic curves; acquiring the adaptation degrees of any cluster to each resource node, and sorting each resource node in descending order of the adaptation degrees; screening resource nodes that meet a preset allowable order position according to the sorting result, obtaining the target scheduling nodes of the cluster, and counting the target scheduling nodes of all clusters.

[0030] In the present embodiment, the resource characteristic curve is a resource occupation curve generated by the resource node in historical full-load operation, and the task characteristic curve is a required resource curve of the sub-task. The adaptation degrees of each resource node to each cluster are obtained through the average value of the correlation coefficients of the resource characteristic curves and the task characteristic curves of all sub-tasks in the cluster, so as to screen a plurality of target scheduling nodes corresponding to each cluster.

[0031] S104, calculating scheduling parameters of each sub-task to the corresponding target scheduling node, and scheduling each sub-task according to the scheduling parameters.

[0032] In some embodiments of the present application, the calculating of the scheduling parameters of each sub-task to the corresponding target scheduling node comprises: determining a resource occupation maximum value of each sub-task according to a resource occupation curve of each sub-task, acquiring real-time residual resources of the target scheduling node of the cluster to which the sub-task belongs, and calculating a difference value between a real-time resource availability and the resource occupation maximum value; determining a resource availability of the target scheduling node according to the difference value between the real-time resource availability and the resource occupation maximum value, acquiring an adaptation degree of the cluster to which the sub-task belongs to the target scheduling node, determining a scheduling parameter of the sub-task to the target scheduling node according to the adaptation degree and the resource availability, and counting the scheduling parameters of all sub-tasks to the corresponding target scheduling node.

[0033] In the present embodiment, the real-time resource availability and the resource occupation maximum value of the target scheduling node corresponding to the sub-task are screened, the resource availability of the corresponding target scheduling node is calculated, and the scheduling parameter of the corresponding sub-task is determined through the resource availability and the adaptation degree of the target scheduling node.

[0034] In some embodiments of the present application, the determining of the scheduling parameter of the sub-task to the target scheduling node according to the adaptation degree and the resource availability comprises: determining the scheduling parameter according to a scheduling parameter calculation formula, wherein the scheduling parameter is,

[0035] wherein, is a scheduling parameter, is an adaptation degree, is a resource availability, , respectively a preset first weight and a preset second weight.

[0036] In the embodiment, the adaptation degree and the resource availability are weighted and summed by the preset first weight and the preset second weight to obtain the scheduling parameter, wherein, .

[0037] In some embodiments of the application, the scheduling of each subtask according to the scheduling parameter comprises: obtaining a preset scheduling threshold, calculating the scheduling parameter of the subtask and the corresponding target scheduling node according to the sorting bit; determining whether the scheduling parameter is greater than the preset scheduling threshold, if the scheduling parameter is greater than the preset scheduling threshold, the subtask is scheduled to the corresponding target scheduling node; if the scheduling parameter is less than or equal to the preset scheduling threshold, the scheduling parameter of the target scheduling node of the next sorting bit and the subtask is continued to be calculated until all subtasks are completed.

[0038] Based on the same technical concept, as shown in Figure 2 The application also provides a real-time task automatic scheduling management system, comprising: an extraction module for obtaining a current task and extracting subtasks of the current task, determining a task running feature set according to the subtasks of the current task; a clustering module for clustering each subtask according to the task running feature set, dividing each subtask into a corresponding cluster; a target module for obtaining a resource feature curve of a current resource node and a task feature curve of each subtask in each cluster, determining a target scheduling node of each cluster according to the resource feature curve and the task feature curve; and a scheduling module for calculating a scheduling parameter of each subtask to the corresponding target scheduling node, scheduling each subtask according to the scheduling parameter.

[0039] By applying the above technical solutions, the application obtains a current task and extracts subtasks of the current task, determines a task running feature set according to the subtasks of the current task; clusters each subtask according to the task running feature set, divides each subtask into a corresponding cluster; obtains a resource feature curve of a current resource node and a task feature curve of each subtask in each cluster, determines a target scheduling node of each cluster according to the resource feature curve and the task feature curve; calculates a scheduling parameter of each subtask to the corresponding target scheduling node, and schedules each subtask according to the scheduling parameter. The application can efficiently and reasonably allocate the resource nodes of the task in real time, improving the resource utilization rate.

[0040] Those skilled in the art can clearly understand the present application through the description of the foregoing embodiments that the present application can be realized by hardware or by means of software and necessary universal hardware platform. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a U disk, a mobile hard disk, etc.), and includes several instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in various implementation scenarios of the present application.

[0041] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacement for some technical features thereof; and these modifications or replacements do not drive the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A real-time automated task scheduling and management method, characterized in that, include: Get the current task and extract its subtasks, and determine the task execution feature set based on the subtasks of the current task; Based on the task execution feature set, each subtask is clustered and assigned to a corresponding cluster. Obtain the resource characteristic curve of the current resource node and the task characteristic curve of the subtasks within each cluster, and determine the target scheduling node for each cluster based on the resource characteristic curve and the task characteristic curve. Calculate the scheduling parameters of each subtask for the corresponding target scheduling node, and schedule each subtask according to the scheduling parameters.

2. The real-time task automated scheduling and management method according to claim 1, characterized in that, The step of determining the task execution feature set based on the subtasks of the current task includes: Obtain historical task execution data, and extract subtasks for each task based on the historical task execution data; Statistically analyze the changes in resource usage of subtasks and plot resource usage curves based on these changes. Determine the task operation characteristics of subtasks based on resource usage curves, and establish a training sample set based on the task operation characteristics of subtasks; A running feature extraction model is established based on the training sample set, and the running feature extraction model is trained to obtain a trained running feature extraction model; Input the subtasks of the current task into the trained runtime feature extraction model to obtain the task runtime feature set.

3. The real-time task automated scheduling and management method according to claim 2, characterized in that, The process of determining the task execution characteristics of subtasks based on resource usage curves includes: Obtain a preset segmentation segment, and divide the resource usage curve of the subtask according to the preset segmentation segment to obtain several resource usage curve segments; Obtain the maximum and minimum values ​​of each resource usage curve segment, calculate the difference between the maximum and minimum values ​​in the resource usage curve segment, and obtain the first task operation characteristics; The average value of each resource usage curve segment is obtained to determine the operating characteristics of the second task. The number of resource usage curve segments whose average value is greater than the preset standard average value is used to obtain the third task operation characteristics; The task operation characteristics of the subtasks are determined based on the first task operation characteristics, the second task operation characteristics, and the third task operation characteristics.

4. The real-time task automated scheduling and management method according to claim 1, characterized in that, The clustering of subtasks based on the task execution feature set includes: A sample dataset is created based on each subtask of the current task, and k initial cluster centers are randomly selected from the sample dataset. Calculate the feature distance from the subtasks in the sample dataset to the initial cluster center, and divide each subtask into its corresponding cluster based on the feature distance from the subtasks in the sample dataset to the initial cluster center; Calculate the mean of the task execution characteristics of subtasks within each cluster, and reselect cluster centers based on the mean of the characteristic distances of subtasks within each cluster; Repeat the above steps until the cluster centers no longer change or the number of iterations reaches the preset maximum number of iterations, and obtain the clustering results of the subtask.

5. The real-time task automated scheduling and management method according to claim 4, characterized in that, The calculation of the feature distance from the subtasks in the sample dataset to the initial cluster centers includes: The feature distance is determined according to the feature distance calculation formula, which is as follows: in, For feature distance, The first task's running characteristics represent the i-th curve segment of the subtask. The first task running feature of the i-th curve segment for cluster centers. The second task's runtime characteristics represent the i-th curve segment of the subtask. The second task operation feature is the i-th curve segment for cluster centers. This is the third task's operational characteristic within the subtask. The third task operation characteristic of cluster centers, The number of curve segments. , , These are the first conversion factor, the second conversion factor, and the third conversion factor, respectively. .

6. The real-time task automated scheduling and management method according to claim 1, characterized in that, The step of determining the target scheduling node for each cluster based on resource characteristic curves and task characteristic curves includes: Calculate the correlation coefficient between the resource characteristic curve and the task characteristic curve, and determine the degree of fit of each resource node to each cluster based on the correlation coefficient between the resource characteristic curve and the task characteristic curve. Obtain the fit degree between any cluster and each resource node, and sort the resource nodes in descending order of fit degree; Based on the sorting results, resource nodes that meet the preset allowable sorting order are selected to obtain the target scheduling nodes of the clusters, and the target scheduling nodes of all clusters are counted.

7. The real-time task automated scheduling and management method according to claim 1, characterized in that, The calculation of scheduling parameters for each subtask relative to the corresponding target scheduling node includes: Determine the maximum resource usage of each subtask based on its resource usage curve, obtain the real-time remaining resources of the target scheduling node of the cluster to which the subtask belongs, and calculate the difference between the real-time resource availability and the maximum resource usage. The resource availability of the target scheduling node is determined based on the difference between the real-time resource availability and the maximum resource usage. The degree of fit between the cluster to which the subtask belongs and the target scheduling node is obtained. The scheduling parameters of the subtask to the target scheduling node are determined based on the degree of fit and the resource availability. The scheduling parameters of all subtasks to the corresponding target scheduling node are statistically analyzed.

8. The real-time task automated scheduling and management method according to claim 7, characterized in that, The step of determining the scheduling parameters of the subtask for the target scheduling node based on the degree of adaptation and resource availability includes: The scheduling parameters are determined according to the formula for calculating the scheduling parameters, and the scheduling parameters are as follows: in, For scheduling parameters, To ensure compatibility, For resource availability, , These are the preset first weight and the preset second weight, respectively.

9. The real-time task automated scheduling and management method according to claim 6, characterized in that, The scheduling of each subtask according to the scheduling parameters includes: Obtain the preset scheduling threshold and calculate the scheduling parameters of subtasks and corresponding target scheduling nodes according to their sorting order; Determine if the scheduling parameters are greater than the preset scheduling threshold. If the scheduling parameters are greater than the preset scheduling threshold, schedule the subtask to the corresponding target scheduling node. If the scheduling parameters are less than or equal to the preset scheduling threshold, the scheduling parameters of the target scheduling node and subtasks in the next sorting position are calculated until all subtasks are scheduled.

10. A real-time task automated scheduling and management system, characterized in that, include: The extraction module is used to obtain the current task and extract its subtasks, and determine the task execution feature set based on the subtasks of the current task; The clustering module is used to cluster each subtask according to the task's running feature set, and to divide each subtask into the corresponding cluster. The target module is used to obtain the resource characteristic curve of the current resource node and the task characteristic curve of the subtasks within each cluster, and to determine the target scheduling node for each cluster based on the resource characteristic curve and the task characteristic curve. The scheduling module is used to calculate the scheduling parameters of each subtask to the corresponding target scheduling node, and to schedule each subtask according to the scheduling parameters.

Images