Workflow scheduling method and device, equipment, storage medium and program product

Abstract

The embodiment of the invention provides a workflow scheduling method and device, equipment, a storage medium and a program product, and is applied to the technical field of cloud computing. The method comprises the steps of obtaining target attribute data corresponding to a target workflow; according to the target attribute data, constructing a multi-target optimization model corresponding to the target workflow; the multi-target optimization model comprises a target hybrid optimization algorithm; determining a target solution set corresponding to the target workflow according to the target attribute data and a multi-target optimization model; and decoding the target solution set to obtain a target workflow scheduling mode corresponding to the target workflow. Therefore, the electronic equipment calculates and optimizes the workflow scheduling based on the multi-objective optimization model, so that the diversity of the workflow scheduling process and the scheduling result can be improved, and meanwhile, the optimization effect of the workflow scheduling can also be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a workflow scheduling method, apparatus, device, storage medium and program product. Background Art

[0002] With the continuous development of Internet technology, the digital transformation of finance and other fields continues to deepen. In online businesses such as cloud services, there is a certain degree of uncertainty in the execution process of workflows, and because the factors affecting workflows are dynamic and diverse, the scheduling and execution of workflows are highly dynamic.

[0003] Since there are many factors affecting workflows, workflow scheduling usually requires optimization for multiple objectives. For the multi-objective optimization problem of workflow scheduling, the relevant technology usually sorts the influencing factors, and then converts the multi-objective optimization process into a single-objective optimization process, and finally obtains a single solution. This workflow scheduling method cannot take into account the multiple influencing factors of the workflow, the optimization process and optimization results are relatively simple, and the optimization effect for workflow scheduling is limited. Summary of the invention

[0004] The embodiments of the present application provide a workflow scheduling method, apparatus, device, storage medium, and program product, which can improve the diversity of workflow scheduling processes and optimization results, and can improve the optimization effect of workflow scheduling.

[0005] In a first aspect, an embodiment of the present application provides a workflow scheduling method, including:

[0006] Obtain target attribute data corresponding to the target workflow;

[0007] According to the target attribute data, a multi-objective optimization model corresponding to the target workflow is constructed; the multi-objective optimization model includes a target hybrid optimization algorithm;

[0008] Determining a target solution set corresponding to the target workflow according to the target attribute data and the multi-objective optimization model;

[0009] The target solution set is decoded to obtain a target workflow scheduling method corresponding to the target workflow.

[0010] In a possible implementation, the target attribute data includes execution task data, computing resource data, execution time data, execution cost data, and load balancing data.

[0011] In a possible implementation, constructing a multi-objective optimization model corresponding to the target workflow according to the target attribute data includes:

[0012] Determine, according to the execution task data, a directed acyclic graph corresponding to the target workflow, and determine an entry task node and an exit task node corresponding to the directed acyclic graph;

[0013] Determining a target workflow scheduling model corresponding to the target workflow according to the computing resource data and the directed acyclic graph;

[0014] A multi-objective optimization model corresponding to the target workflow scheduling model is constructed according to the execution time data, the execution cost data, the load balancing data and the target workflow scheduling model.

[0015] In a possible implementation, determining the target solution set corresponding to the target workflow according to the target attribute data and the multi-objective optimization model includes:

[0016] Initializing the multi-objective optimization model according to the execution task data corresponding to the target workflow;

[0017] According to the computing resource data, encoding processing is performed on the directed acyclic graph corresponding to the target workflow and the mapping relationship between computing resources to obtain an initial feasible solution corresponding to the multi-objective optimization model;

[0018] The initial feasible solution is calculated and processed by the target hybrid optimization algorithm in the multi-objective optimization model to obtain a target solution set corresponding to the target workflow.

[0019] In a possible implementation, the target hybrid optimization algorithm is a genetic bacterial foraging optimization algorithm; the genetic bacterial foraging optimization algorithm includes a genetic optimization algorithm and a bacterial foraging optimization algorithm; the target hybrid optimization algorithm in the multi-objective optimization model is used to calculate the initial feasible solution to obtain a target solution set corresponding to the target workflow, including:

[0020] According to the genetic optimization algorithm, the initial feasible solution is iterated to obtain an intermediate solution set;

[0021] When the intermediate solution set meets the target switching condition, the intermediate solution set is iteratively calculated according to the bacterial foraging optimization algorithm to obtain the target solution set corresponding to the target workflow.

[0022] In a possible implementation, the genetic optimization algorithm includes a selection operator based on a Pareto dominance relationship, an adaptive crossover operator, and an adaptive mutation operator;

[0023] The target switching condition includes that the minimum average evolution rate of the population and the individual fitness variance of the population meet the preset conditions, or the number of iterations of the genetic optimization algorithm reaches the maximum number of iterations;

[0024] The bacterial foraging optimization algorithm includes a trending operation using an adaptive trending step size based on a step size adjustment factor and a dispersing operation based on grid division.

[0025] In a second aspect, an embodiment of the present application provides a workflow scheduling device, including:

[0026] An acquisition module is used to acquire target attribute data corresponding to a target workflow;

[0027] A construction module, used to construct a multi-objective optimization model corresponding to the target workflow according to the target attribute data; the multi-objective optimization model includes a hybrid optimization algorithm;

[0028] A determination module, used to determine a target solution set corresponding to the target workflow according to the target attribute data and the multi-objective optimization model;

[0029] The decoding module is used to decode the target solution set to obtain the target workflow scheduling mode corresponding to the target workflow.

[0030] In a possible implementation, the target attribute data includes execution task data, computing resource data, execution time data, execution cost data, and load balancing data.

[0031] In a possible implementation manner, the building block is specifically used to:

[0032] Determine, according to the execution task data, a directed acyclic graph corresponding to the target workflow, and determine an entry task node and an exit task node corresponding to the directed acyclic graph;

[0033] Determining a target workflow scheduling model corresponding to the target workflow according to the computing resource data and the directed acyclic graph;

[0034] A multi-objective optimization model corresponding to the target workflow scheduling model is constructed according to the execution time data, the execution cost data, the load balancing data and the target workflow scheduling model.

[0035] In a possible implementation manner, the determining module is specifically configured to:

[0036] Initializing the multi-objective optimization model according to the execution task data corresponding to the target workflow;

[0037] According to the computing resource data, encoding processing is performed on the directed acyclic graph corresponding to the target workflow and the mapping relationship between computing resources to obtain an initial feasible solution corresponding to the multi-objective optimization model;

[0038] The initial feasible solution is calculated and processed by the target hybrid optimization algorithm in the multi-objective optimization model to obtain a target solution set corresponding to the target workflow.

[0039] In a possible implementation, the target hybrid optimization algorithm is a genetic bacterial foraging optimization algorithm; the genetic bacterial foraging optimization algorithm includes a genetic optimization algorithm and a bacterial foraging optimization algorithm; the determination module is specifically used to:

[0040] According to the genetic optimization algorithm, the initial feasible solution is iterated to obtain an intermediate solution set;

[0041] When the intermediate solution set meets the target switching condition, the intermediate solution set is iteratively calculated according to the bacterial foraging optimization algorithm to obtain the target solution set corresponding to the target workflow.

[0042] In a possible implementation, the genetic optimization algorithm includes a selection operator based on a Pareto dominance relationship, an adaptive crossover operator, and an adaptive mutation operator;

[0043] The target switching condition includes that the minimum average evolution rate of the population and the individual fitness variance of the population meet the preset conditions, or the number of iterations of the genetic optimization algorithm reaches the maximum number of iterations;

[0044] The bacterial foraging optimization algorithm includes a trending operation using an adaptive trending step size based on a step size adjustment factor and a dispersing operation based on grid division.

[0045] In a third aspect, an embodiment of the present application provides a workflow scheduling device, including: a memory, a processor;

[0046] The memory stores computer-executable instructions;

[0047] The processor executes the computer-executable instructions stored in the memory, so that the processor executes the workflow scheduling method as described in any one of the first aspects above.

[0048] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement the workflow scheduling method as described in any one of the first aspects above.

[0049] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the workflow scheduling method as described in any one of the first aspects above.

[0050] The workflow scheduling method, device, equipment, storage medium and program product provided in the embodiment of the present application obtain the target attribute data corresponding to the target workflow; construct a multi-objective optimization model corresponding to the target workflow according to the target attribute data; the multi-objective optimization model includes a target hybrid optimization algorithm; determine the target solution set corresponding to the target workflow according to the target attribute data and the multi-objective optimization model; decode the target solution set to obtain the target workflow scheduling method corresponding to the target workflow. In the present application, the electronic device obtains the target attribute data of the target workflow, and then constructs the multi-objective optimization model corresponding to the target workflow according to the target attribute data; then determines the target solution set of the target workflow through the target hybrid optimization algorithm and the target attribute data in the multi-objective optimization model, and finally decodes the target solution set to obtain the target workflow scheduling method corresponding to the target workflow. In this way, the electronic device calculates and optimizes the workflow scheduling based on the multi-objective optimization model, which can improve the diversity of the workflow scheduling process and scheduling results, and also can improve the optimization effect of workflow scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0052] Figure 1 A flowchart of a workflow scheduling method provided in an embodiment of the present application;

[0053] Figure 2 A flowchart of another workflow scheduling method provided in an embodiment of the present application;

[0054] Figure 3 A schematic diagram of the structure of a workflow scheduling device provided in an embodiment of the present application;

[0055] Figure 4 A schematic diagram of the structure of a workflow scheduling device provided in an embodiment of the present application.

[0056] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0057] In order to enable those skilled in the art to better understand the technical solution of the present application, the present application is further described in detail below in conjunction with the drawings and embodiments. It should be understood that the specific embodiments and drawings described herein are only used to explain the present application, and are not intended to limit the present application.

[0058] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards, and provide corresponding operation portals for users to choose to authorize or refuse. In the technical solution of this application, the collection, storage, use, processing, transmission, provision and disclosure of user data information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good customs.

[0059] It should be noted that in the embodiments of the present application, certain software, components, models and other existing solutions in the industry may be mentioned, and they should be regarded as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of the present application, but it does not mean that the applicant has or will necessarily use the solution.

[0060] It should be noted that the workflow scheduling method, apparatus, equipment, storage medium and program product of the present application can be used in the field of cloud computing technology, and can also be used in any field outside the field of cloud computing technology. The present application does not limit the specific application field of the workflow scheduling method, apparatus, equipment, storage medium and program product.

[0061] In cloud services, there is a certain degree of uncertainty in the execution of workflows. For example, iterations and repeated operations during execution will cause uncertainty in the total amount of tasks and processes. At the same time, due to the timeliness of load information, the number of available resources and loads for workflows are changing at any time, which is inconsistent with the static allocation. In addition, due to the dynamic diversity of influencing factors, such as the execution capacity and fluctuation of computing resources, and the unquantifiable difficulty of task execution, the scheduling and execution of workflows are dynamic. The global optimal static algorithm in the initial design is too ideal and still not suitable in special circumstances. Therefore, dynamic scheduling algorithms are usually used for scheduling workflows.

[0062] With the continuous development of Internet technology, cloud computing integrates large-scale distributed computing resources through the network and has developed rapidly in many industries and applications. Compared with traditional processing and computing modes, using cloud computing services to complete workflow scheduling and execution has significant advantages in performance and efficiency. The dynamic scheduling of workflows in a cloud computing environment needs to comprehensively consider multiple factors such as user service quality and cloud service provider costs. The workflow scheduling problem can be converted into a multi-objective optimization problem, such as a non-deterministic polynomial (NP-hard) problem with high difficulty.

[0063] Multi-objective optimization problems refer to optimization problems with multiple objective functions under constraints in a system or project. The reason why such problems are difficult to solve is that the multiple sub-objectives of optimization are generally interconnected and constrained by each other. The key to solving them lies in maintaining diversity and the rationality of strategies. In related technologies, multi-objective optimization problems for workflow scheduling are usually solved by sorting the influencing factors according to the degree of association or preference and assigning weights to convert them into single-objective optimization problems, integrating the objective functions, and finally obtaining the optimal solution or near-optimal solution. The weighting and variable weighting of this multi-objective optimization method are somewhat subjective, ignoring the different dimensions and complex topological structures between multiple sub-objectives, and the optimization progress of each objective is not operable. In the end, only one solution under weight distribution can be given. The optimization process and optimization results are relatively simple, and the optimization effect on the workflow is limited.

[0064] In order to solve the above problems, the workflow scheduling method, device, equipment, storage medium and program product in the present application, the electronic device obtains the target attribute data corresponding to the target workflow; according to the target attribute data, a multi-objective optimization model corresponding to the target workflow is constructed; the multi-objective optimization model includes a target hybrid optimization algorithm; according to the target attribute data and the multi-objective optimization model, the target solution set corresponding to the target workflow is determined; the target solution set is decoded to obtain the target workflow scheduling method corresponding to the target workflow. In the present application, the electronic device obtains the target attribute data of the target workflow, and then constructs the multi-objective optimization model corresponding to the target workflow according to the target attribute data; then the target solution set of the target workflow is determined by the target hybrid optimization algorithm and the target attribute data in the multi-objective optimization model, and finally the target solution set is decoded to obtain the target workflow scheduling method corresponding to the target workflow. In this way, the electronic device calculates and optimizes the workflow scheduling based on the multi-objective optimization model, which can improve the diversity of the workflow scheduling process and scheduling results, and also improve the optimization effect of the workflow scheduling.

[0065] The scheme shown in the present application is described in detail below through specific embodiments. It should be noted that the following embodiments can exist independently or in combination with each other, and the same or similar contents will not be described repeatedly in different embodiments.

[0066] Figure 1 A flowchart of a workflow scheduling method provided in an embodiment of the present application. Figure 1 , the workflow scheduling method may include:

[0067] S101. Obtain target attribute data corresponding to a target workflow.

[0068] The execution subject of the embodiment of the present application can be an electronic device, such as a mobile phone, a computer, and a cloud server, etc., and can be a workflow scheduling system installed in the electronic device, or a workflow scheduling device set in the electronic device. The workflow scheduling device can be implemented by software, or by a combination of software and hardware. For ease of understanding, the following description is taken as an example of an execution subject being an electronic device.

[0069] In an embodiment of the present application, the target workflow may refer to a target workflow that needs to be scheduled and optimized. The target attribute data may refer to various attribute data corresponding to the target workflow, which may specifically include execution task data, computing resource data, execution time data, execution cost data, and load balancing data. Among them, the execution task data may reflect the execution status and specific execution status of each task in the target workflow, which may specifically include the dependency relationship and hierarchical structure between tasks in the target workflow, as well as the start time, end time, execution status (e.g., success, failure, executing, etc.), specific operation instructions, etc. of each task. The computing resource data may reflect the computing resources required during the execution of the target workflow, and the computing resource data may specifically include cloud computing resources and virtual machine computing resources in an Infrastructure as a Service (IaaS) environment.

[0070] Execution time data may refer to the execution completion time of each task of the target workflow, and may specifically include the execution time of each task in the target workflow, the total execution completion time of the target workflow, the waiting time between each task, etc. Execution cost data may reflect the various costs incurred during the execution of the target workflow, and may specifically include computing costs, storage costs, network costs, etc. Load balancing data may reflect the load conditions of each node or server during the execution of the target workflow, and may specifically include the load balancing conditions of heterogeneous virtual machine resources or virtual machine available resources. Of course, it should be noted that the target attribute data may also include other types of attribute data, which can be flexibly set in actual application, and the embodiments of the present application are not limited to this.

[0071] Specifically, when scheduling optimization is required for a target workflow, the electronic device may first obtain various target attribute data of the target workflow, and then perform model building and solution set calculation based on the target attribute data.

[0072] S102. Construct a multi-objective optimization model corresponding to the target workflow according to the target attribute data; the multi-objective optimization model includes a target hybrid optimization algorithm.

[0073] In the embodiment of the present application, the multi-objective optimization mode may refer to an optimization model corresponding to the target workflow and used for solution set calculation. The multi-objective optimization model may include a target hybrid optimization algorithm, which may be used to calculate the multi-objective optimization process of the target workflow scheduling, and realize the rapid solution of the target workflow scheduling. The target hybrid optimization algorithm may specifically include a genetic bacterial foraging optimization algorithm (GBFOA), a genetic simulated annealing algorithm (GSA), and a particle swarm genetic algorithm (PSGA), etc. Of course, other types of optimization algorithms may also be included, which can be flexibly selected based on actual needs, and the embodiment of the present application does not limit this.

[0074] Specifically, after obtaining the target attribute data of the target workflow, the electronic device can first determine the workflow scheduling model corresponding to the target workflow, and then determine the multi-objective optimization model corresponding to the target workflow based on the workflow scheduling model and the specific optimization target data in the target attribute data, such as execution completion time, execution cost, and load balancing. In this way, the electronic device can construct a corresponding multi-objective optimization model according to the actual situation of the target workflow, so that the result of the multi-objective optimization can meet the actual needs of the target workflow scheduling and improve the optimization effect of the target workflow scheduling.

[0075] S103: Determine a target solution set corresponding to the target workflow according to the target attribute data and the multi-objective optimization model.

[0076] In the embodiment of the present application, the target solution set may refer to the solution set of the multi-objective optimization problem output by the multi-objective optimization model and corresponding to the target workflow scheduling, and specifically may refer to the Pareto solution set, etc. The target solution set may include multiple target feasible solutions, each target feasible solution corresponding to a target workflow scheduling method.

[0077] In this step, after constructing the multi-objective optimization model corresponding to the target workflow, the electronic device can determine the input data of the multi-objective optimization model based on the target attribute data. The input data can refer to the directed acyclic graph (DAG) corresponding to the target workflow and the computing resources corresponding to the target workflow, such as the virtual machine resource set and virtual machine type set corresponding to the target workflow, and the computing resources can be determined based on the computing resources corresponding to the target workflow. Afterwards, the target hybrid optimization algorithm in the multi-objective optimization model can perform iterative calculations based on the input data to determine the target solution set corresponding to the target workflow. In this way, the electronic device can directly solve and calculate the target workflow scheduling based on the multi-objective optimization model, without converting the multi-objective optimization problem into a single-objective optimization problem, and can ensure the interaction of various factors in the target workflow scheduling, can ensure the rationality of the workflow scheduling, improve the optimization effect, and can obtain a target solution set including multiple target feasible solutions, which can improve the diversity and uniformity of the workflow scheduling optimization.

[0078] S104: Decode the target solution set to obtain a target workflow scheduling method corresponding to the target workflow.

[0079] In the embodiment of the present application, the target workflow scheduling method may refer to the target scheduling strategy corresponding to the target workflow. Specifically, since the target hybrid optimization algorithm in the multi-objective optimization model may adopt an iterative calculation method for general scenarios, the target solution set outputted by it cannot be directly applied to the scheduling scenario of the target workflow in cloud computing. The electronic device can decode the target solution set based on a preset decoding method to obtain the target workflow scheduling method corresponding to the target workflow, thereby realizing the optimization processing for the scheduling of the target workflow.

[0080] The workflow scheduling method provided by the embodiment of the present application, the electronic device obtains the target attribute data corresponding to the target workflow; according to the target attribute data, a multi-objective optimization model corresponding to the target workflow is constructed; the multi-objective optimization model includes a target hybrid optimization algorithm; according to the target attribute data and the multi-objective optimization model, the target solution set corresponding to the target workflow is determined; the target solution set is decoded to obtain the target workflow scheduling method corresponding to the target workflow. In the present application, the electronic device obtains the target attribute data of the target workflow, and then constructs the multi-objective optimization model corresponding to the target workflow according to the target attribute data; then the target solution set of the target workflow is determined by the target hybrid optimization algorithm and the target attribute data in the multi-objective optimization model, and finally the target solution set is decoded to obtain the target workflow scheduling method corresponding to the target workflow. In this way, the electronic device calculates and optimizes the workflow scheduling based on the multi-objective optimization model, which can improve the diversity of the workflow scheduling process and scheduling results, and also can improve the optimization effect of the workflow scheduling.

[0081] Based on the above embodiments, Figure 2 A flowchart of another workflow scheduling method provided in an embodiment of the present application. Figure 2 , the workflow scheduling method may include:

[0082] S201. Obtain target attribute data corresponding to the target workflow; the target attribute data includes execution task data, computing resource data, execution time data, execution cost data, and load balancing data.

[0083] In an embodiment of the present application, when the electronic device schedules and optimizes the target workflow, it can obtain the target attribute data corresponding to the target workflow. Specifically, it can be obtained through the task management system, cloud platform monitoring service and workflow configuration data corresponding to the target workflow. The embodiment of the present application does not limit the specific method of obtaining the target attribute data.

[0084] S202: Determine a directed acyclic graph corresponding to the target workflow according to the execution task data, and determine an entry task node and an exit task node corresponding to the directed acyclic graph.

[0085] In the embodiment of the present application, the directed acyclic graph may refer to a directed graph without loops corresponding to the target workflow, which can characterize the hierarchical structure and task dependencies of the target workflow. The entry task node may refer to an entry node in the directed acyclic graph, corresponding to the initial task node in the target workflow; the exit task node may refer to an exit node in the directed acyclic graph, corresponding to the termination task node in the target workflow.

[0086] Specifically, after acquiring the target attribute data of the target workflow, the electronic device can abstract the target workflow into a directed acyclic graph according to the execution task data in the target attribute data, and configure an entry task node and an exit task node for the directed acyclic graph.

[0087] S203: Determine a target workflow scheduling model corresponding to the target workflow according to the computing resource data and the directed acyclic graph.

[0088] S204: Construct a multi-objective optimization model corresponding to the target workflow scheduling model according to the execution time data, the execution cost data, the load balancing data and the target workflow scheduling model.

[0089] In the embodiment of the present application, the target workflow scheduling model may refer to a mathematical model corresponding to the target workflow, which can reflect the task dependencies and computing resources of the target workflow. Specifically, after obtaining the directed acyclic graph corresponding to the target workflow, the electronic device can establish a target workflow scheduling model corresponding to the target workflow according to the directed acyclic graph and the computing resource data corresponding to the target workflow. Then, the electronic device can construct a multi-objective optimization model corresponding to the target workflow based on the target workflow scheduling model and the execution time data, execution cost data and load balancing data in the target attribute data. In this way, the electronic device establishes a target workflow scheduling model based on the execution task data and the computing resource data, and then constructs a multi-objective optimization model based on the target workflow scheduling model and the execution time data, execution cost data and load balancing data, which can improve the rationality of the target workflow scheduling optimization, and can improve the user service quality indicators while ensuring the normal execution of the target workflow, such as shortening the execution time and reducing the execution cost.

[0090] S205: Initialize the multi-objective optimization model according to the execution task data corresponding to the target workflow.

[0091] In an embodiment of the present application, after constructing a multi-objective optimization model corresponding to the target workflow, the electronic device can initialize the multi-objective optimization model according to the execution task data corresponding to the target workflow. Specifically, the electronic device can use a directed acyclic graph determined based on the execution task data as part of the input data, and initialize the multi-objective optimization model based on the hierarchical structure of the workflow reflected in the directed acyclic graph. Exemplarily, when the target hybrid optimization algorithm is a genetic bacterial foraging optimization algorithm, the electronic device can perform a hierarchical traversal of the workflow hierarchical structure based on the directed acyclic graph and perform population initialization operations; at the same time, the electronic device can determine the initialization priority of the target workflow under the available virtual machine resources, and can subsequently update the population queue based on the initialization priority, so as to ensure that the dependencies between the tasks of the target workflow are not destroyed.

[0092] S206. According to the computing resource data, the directed acyclic graph corresponding to the target workflow and the mapping relationship between the computing resources are encoded to obtain an initial feasible solution corresponding to the multi-objective optimization model.

[0093] In an embodiment of the present application, the initial feasible solution may refer to the initial solution corresponding to the target hybrid optimization algorithm in the multi-objective optimization model. Specifically, the electronic device may first determine the mapping relationship between the target workflow task represented by the directed acyclic graph and the computing resources corresponding to the target workflow based on the directed acyclic graph and the computing resource data, and then the electronic device may uniformly encode the mapping relationship based on a preset encoding method to obtain the initial feasible solution of the multi-objective optimization model. The preset encoding method corresponds to the preset decoding method, and specifically may refer to the encoding and decoding method of the target hybrid optimization algorithm applicable to the workflow scheduling scenario, which is not limited in the embodiment of the present application.

[0094] S207, calculating and processing the initial feasible solution through the target hybrid optimization algorithm in the multi-objective optimization model to obtain a target solution set corresponding to the target workflow.

[0095] In an embodiment of the present application, after determining the initial feasible solution, the electronic device can iteratively process the initial feasible solution through the target hybrid optimization algorithm in the target optimization model, and finally the target hybrid optimization algorithm can output the target solution set corresponding to the target workflow.

[0096] In an embodiment of the present application, the target hybrid optimization algorithm may refer to a genetic bacterial foraging optimization algorithm; the genetic bacterial foraging optimization algorithm includes a genetic optimization algorithm and a bacterial foraging optimization algorithm.

[0097] In related technologies, genetic algorithms (GA) adaptively search for optimal solutions or near-optimal solutions by simulating the biological evolution process of natural selection and genetics. Although different genetic operators and encoding and decoding schemes correspond to different scenarios, the operations of the basic genetic algorithm (Simple Genetic Algorithms, SGA) are selection, crossover and mutation, which is the basic framework for improving genetic evolution algorithms.

[0098] Bacterial Foraging Optimization Algorithm (BFOA) can simulate the basic law of growth evolution of human Escherichia coli colonies, and achieve optimization through three behaviors: chemotaxis, replication, and elimination-dispersal. The chemotaxis operation originates from the phenomenon that bacteria gather in nutrient-rich areas. The corresponding problem solving process is the process of searching in local areas. The chemotaxis operation is divided into two steps: flipping and advancing. Flipping determines the forward direction of bacteria. Advancing is the moving step after flipping is completed, and the improvement of fitness value is continuously compared during the forward process until the predetermined moving step threshold is reached or the fitness is no longer improved. The chemotaxis operation defines the local search ability and detail of bacteria. After reaching the critical value of the chemotaxis, bacteria with strong foraging ability will reproduce themselves. The replication operation simulates this behavior, retains some bacteria with high fitness values ​​to guide the replication of bacteria, and keeps the bacterial scale unchanged, thereby improving convergence and computational efficiency, and further accelerating the search speed of bacteria. For complex optimization problems, chemotaxis and replication cannot prevent bacteria from falling into the local optimal phenomenon. By introducing the dispersal operation, the bacteria can be dispersed to any position in the search space with a certain probability after completing a certain number of replications. The dispersal operation enables the bacteria to have random search capabilities, while enhancing the stability and diversity of the algorithm. By jumping out of the local optimal solution, repeated iterations are avoided, and the probability of premature convergence is reduced.

[0099] The genetic algorithm is based on the biological evolution process. It has good fast global search capabilities and can quickly obtain feasible solutions in the solution space. When the calculation accuracy is not high, it has good convergence, thus avoiding the trap of rapid descent of the local optimal solution. However, the local search ability of the genetic algorithm is poor. Therefore, with the rapid increase of the number of iterations of the genetic population, the algorithm solution speed slows down, and the efficiency of the solution search is reduced in the later stage. The genetic information in the parent population and the offspring population in the genetic algorithm is not fully fed back and utilized, which is prone to premature convergence when dealing with complex large-scale computing problems. The bacterial foraging algorithm has a strong local search ability. The chemotaxis operation can strengthen the interaction of feedback information between populations, and the dispersal operation can maintain the diversity of the population. When dealing with multi-objective optimization problems, it can ensure that the feasible solutions in the Pareto solution set are evenly distributed. However, the bacterial foraging algorithm has a strong dependence on the initial population and insufficient global search ability. Its time complexity is high when dealing with complex problems.

[0100] The target hybrid optimization algorithm in the embodiment of the present application can adopt a genetic bacterial foraging optimization algorithm, including a genetic optimization algorithm after optimizing the genetic algorithm and a bacterial foraging optimization algorithm after optimizing the bacterial foraging algorithm. In the embodiment of the present application, the characteristics of the genetic optimization algorithm and the bacterial foraging optimization algorithm can be combined to improve the optimization effect of the two under multi-objective optimization problems, and at the same time, an adaptive mechanism is added to improve the genetic operator and the bacterial foraging optimization operator, thereby improving the computational efficiency and algorithm convergence speed of the genetic algorithm and the bacterial foraging algorithm. In addition, in the embodiment of the present application, for the workflow scheduling scenario under cloud computing, a preset encoding method and a preset decoding method are used to implement the population encoding and decoding scheme and the initialization of the population, and a multi-objective fitness function is determined according to the execution time, execution cost, and load balancing to evaluate the individual fitness. By setting a non-inferior solution buffer pool externally to store the non-inferior solutions generated in the algorithm process, the optimization effect of workflow scheduling can be improved.

[0101] Specifically, in the genetic optimization algorithm, after the electronic device determines the problem decision variables and related constraints to establish the solution space and uses the optimization objective function to establish a multi-objective optimization model, the genetic optimization algorithm determines the search space based on the chromosome (i.e., initial feasible solution) encoding scheme and conversion method of the specific scenario of the target workflow, and in the process of iteratively searching for feasible solutions, the individuals in each evolving generation are measured according to the fitness function value; and selection operations are used to perform combined crossover and mutation operations on them to generate a new generation of population to represent the solution set, and finally the optimal individual in the last generation of population is decoded to obtain an approximate optimal solution or an optimal solution.

[0102] Thus, in the embodiment of the present application, the electronic device performs multi-objective optimization workflow scheduling through a target hybrid optimization algorithm based on a genetic optimization algorithm and a bacterial foraging optimization algorithm. The target hybrid optimization algorithm effectively connects the genetic optimization algorithm and the bacterial foraging optimization algorithm, and improves the intermediate operators of the two respectively, thereby combining the global search capability of the genetic optimization algorithm in the early stage of the solution and the good local optimization capability and better solution accuracy of the bacterial foraging optimization algorithm, thereby avoiding the shortcomings of a single algorithm in the solution process. Combining the optimization of the workflow scheduling process with respect to workflow execution time, execution cost and load balancing can ensure the balance of workflow scheduling.

[0103] In a possible implementation, step S207 may be implemented in the following manner:

[0104] According to the genetic optimization algorithm, the initial feasible solution is iteratively processed to obtain the intermediate solution set; when the intermediate solution set meets the target switching condition, the intermediate solution set is iteratively calculated according to the bacterial foraging optimization algorithm to obtain the target solution set corresponding to the target workflow.

[0105] In an embodiment of the present application, the intermediate solution set may refer to the solution set output by the genetic optimization algorithm. The target switching condition may refer to a pre-set algorithm switching condition. Specifically, after obtaining the initial feasible solution, the electronic device may first iteratively process the initial feasible solution through the genetic optimization algorithm to obtain an intermediate solution set. If the intermediate solution set meets the target switching condition, the electronic device may iteratively calculate the intermediate solution set based on the bacterial foraging optimization algorithm to obtain the target solution set corresponding to the target workflow.

[0106] In the embodiment of the present application, the target hybrid optimization algorithm first initializes the population based on the execution task data of the target workflow, specifically based on the directed acyclic graph corresponding to the execution task data, combined with the workflow hierarchical structure reflected by the directed acyclic graph; then encodes the directed acyclic graph corresponding to the target workflow and the mapping relationship between computing resources to generate an initial feasible solution. In this way, the target hybrid optimization algorithm in the embodiment of the present application adopts a preset encoding method and a preset decoding method from the perspective of cloud computing, and determines a normalized fitness evaluation function based on the Pareto dominance relationship. The population initialization method based on the workflow hierarchical structure is also more in line with the scenario of workflow task scheduling, and can meet the actual needs of workflow scheduling.

[0107] In the embodiment of the present application, after the initialization work is completed, the electronic device uses a genetic optimization algorithm in the first stage to calculate and process the initial feasible solution. The initial feasible solution at this time can also be called a chromosome, and the solution set can be called a population. The electronic device iterates and optimizes under the target switching condition through three genetic operators: a selection operator based on the Pareto dominance relationship, an adaptive crossover operator, and an adaptive mutation operator, and continuously increases the chromosome fitness value during the evolution process. In each iteration, the better chromosomes in the population are retained according to the elite preservation strategy, and the remaining chromosomes are eliminated. The target switching condition includes that the minimum average evolution rate of the population and the individual fitness variance of the population meet the preset conditions, or the number of iterations of the genetic optimization algorithm reaches the maximum number of iterations. In this way, the target switching condition is based on the stability of the solution set, and comprehensively considers factors such as the number of iterations, the minimum average evolution rate, and the individual fitness variance of the population, so that the algorithm switching is more reasonable.

[0108] After the intermediate solution set meets the target switching condition, the electronic device can iteratively process the intermediate solution set through the bacterial foraging optimization algorithm in the second stage. Specifically, after the bacterial foraging optimization algorithm is initialized, it further performs local search through improved trending, replication and dispersal operations. In this process, the maintenance of population diversity is considered to add non-inferior solutions to the non-inferior solution buffer pool, and finally generate a Pareto optimal solution set, which is the target solution set. Exemplarily, the main steps of the target hybrid optimization algorithm are given below:

[0109] S1. Obtain input data and initialize the population to determine an initial feasible solution. The input data is a directed acyclic graph corresponding to the target workflow, a set of virtual machine resources, and a set of virtual machine types.

[0110] If S2, the number of iterations, the minimum average evolution rate, and the individual fitness variance of the population do not meet the target switching conditions, repeat steps S3 to S8.

[0111] S3. Generate a multi-objective dominance relationship performance matrix based on the current population.

[0112] S4. Calculate the selection probability according to the selection operator based on the Pareto dominance relationship and select individuals.

[0113] S5. Calculate the crossover probability according to the adaptive crossover operator and perform a crossover operation on the selected individuals.

[0114] S6. Calculate the mutation probability according to the adaptive mutation operator and apply it to the population after crossover.

[0115] S7. Integrate the sub-population after the genetic optimization algorithm operation and the current population, and iteratively generate a new population according to the elite retention strategy.

[0116] S8: Determine the target switching condition. If the condition is not met, go to step S2. Otherwise, go to step S9.

[0117] S9. Initialize the bacterial population according to the population in the genetic optimization algorithm stage (i.e., the intermediate solution set).

[0118] S10: Repeat steps S11 to S17 within the maximum number of dispersal operations.

[0119] S11: Repeat steps S12 to S16 within the maximum number of copy operations.

[0120] S12: Repeat steps to S13 within the maximum number of trend operations.

[0121] S13. Update the distribution positions of individual bacteria according to the adaptive trend step size.

[0122] S14. Divide the sub-population according to the multi-objective fitness function.

[0123] S15. Based on the genetic crossover operation, individuals are selected from the sub-population for replication operation.

[0124] S16. Update the non-inferior solution buffer pool.

[0125] S17: Determine whether the last operation of the dispersal operation has been reached, and do not perform the dispersal operation for the last time in the life cycle.

[0126] S18. Output the non-inferior solution buffer pool to the Pareto optimal solution set to obtain the target solution set.

[0127] In the embodiment of the present application, the selection operator based on the Pareto dominance relationship in the genetic operator adopts different selection strategies in the non-inferior solution set and the inferior solution set respectively, and unifies the selection probability, which can maintain the diversity of the population to a certain extent. The crossover operator, mutation operator and the tendency operation and dispersal operation in the bacterial foraging optimization algorithm are all added with an adaptive mechanism, which can dynamically adapt according to the changes in the population evolution environment and the degree of evolution. The replication operation combined with the genetic algorithm crossover operation further strengthens the individual communication in the solution space in the local search of the algorithm in the later stage. In addition, the non-inferior solution buffer pool is used in the genetic bacterial foraging optimization algorithm to continuously update the determined non-inferior solutions in the algorithm's later optimization process, thereby ensuring the performance of the non-dominated solution finally obtained; the algorithm switching transition of the target hybrid optimization algorithm comprehensively considers the maximum evolutionary iteration number of the population (i.e., the maximum number of iterations) while combining the population fitness variance change and the evolution rate state, thereby improving the overall solution efficiency of the algorithm.

[0128] In an embodiment of the present application, the electronic device adopts a genetic bacterial foraging optimization algorithm based on a genetic optimization algorithm and a bacterial foraging optimization algorithm as a target hybrid optimization algorithm, through a unified preset encoding method and a preset decoding method for the target solution set, and uses a fitness function to determine the performance based on the dominance relationship between individuals and the objective function, thereby ensuring that individuals with good fitness have a greater chance of being retained.

[0129] In a possible implementation, the genetic optimization algorithm includes a selection operator based on a Pareto dominance relationship, an adaptive crossover operator, and an adaptive mutation operator;

[0130] The target switching conditions include that the minimum average evolution rate of the population and the individual fitness variance of the population meet the preset conditions, or the number of iterations of the genetic optimization algorithm reaches the maximum number of iterations;

[0131] The bacterial foraging optimization algorithm includes a tendency operation with an adaptive tendency step size based on a step size adjustment factor and a dispersal operation based on grid division.

[0132] In the embodiment of the present application, the electronic device improves the genetic algorithm and the bacterial foraging algorithm for workflow scheduling and multi-objective optimization scenarios, and obtains the genetic optimization algorithm and the bacterial foraging optimization algorithm. Specifically, for the selection operator based on the Pareto dominance relationship in the genetic optimization algorithm, the population is divided into a non-inferior solution set and an inferior solution set according to the Pareto dominance relationship. In the non-inferior solution set, the probability of individual selection is determined according to the roulette method based on the dominance number of feasible solutions, and in the inferior solution set, the selection probability is determined based on the variance of the fitness value, thereby ensuring the diversity and uniformity of the selection of feasible solutions and improving the search speed of the algorithm. For the crossover operator and mutation operator in the genetic optimization algorithm, on the one hand, an adaptive mechanism combining evolutionary algebra and fitness value is added, and on the other hand, the task dependency constraints in the operator operation are controlled at the same time. For the trend operation in the bacterial foraging optimization algorithm, an adaptive trend step size with a step size adjustment factor is adopted; the grid partitioning method is cited in the dispersal operation to ensure the diversity and uniformity of the solution space distribution. In order to connect and switch the genetic optimization algorithm and the bacterial foraging optimization algorithm in the target hybrid optimization algorithm, the electronic device adopts a target switching condition, which is based on the stability of the solution set. While considering the maximum evolutionary iteration number of the population, it can combine the changes in the population fitness variance and the evolution rate state to achieve a reasonable connection between the genetic optimization algorithm and the bacterial foraging optimization algorithm.

[0133] It should be noted that the target hybrid optimization algorithm in the embodiment of the present application may also adopt other algorithms, and there may be other corresponding implementation processes, which is not limited in the embodiment of the present application.

[0134] S208: Decode the target solution set to obtain a target workflow scheduling method corresponding to the target workflow.

[0135] In an embodiment of the present application, after the electronic device obtains the target solution set based on the multi-objective optimization model, it can decode the target solution set to obtain a target workflow scheduling method corresponding to the target workflow. This can improve the diversity of the workflow scheduling process and scheduling results, and can also improve the optimization effect of the workflow scheduling.

[0136] In the embodiment of the present application, the electronic device establishes a target workflow scheduling model and a multi-objective optimization model based on the target attribute data of the target workflow, and at the same time, according to the target hybrid optimization algorithm, it can improve the convergence speed of feasible solution calculation, and at the same time, it can improve the uniformity and diversity of the target solution set distribution. Through experimental comparison of the multi-objective optimization results, algorithm performance and the influence of the virtual machine resource scale, the workflow scheduling method based on the target hybrid optimization algorithm in the embodiment of the present application can obtain a relatively good target solution set while optimizing multiple objectives, and has improved the results of each optimization target compared with other existing algorithms, and the algorithm performance is relatively high.

[0137] Specifically, the electronic device in the embodiment of the present application can conduct simulation comparison experiments on the target hybrid optimization algorithm and the multi-objective optimization algorithm in the related technology through the experimental simulation environment platform and parameter settings. Specifically, the average operation results of the optimization objectives of the algorithm under workflow tasks of different scales and structures can be compared; the calculation results of the Pareto solution set obtained by the algorithm under the algorithm performance evaluation index and the existing differences can also be compared; the optimization degree of the scheduling scheme obtained by the algorithm under different available virtual machine resource scales can also be compared.

[0138] In this way, the electronic device conducts simulation experiments through the integrated tool framework to compare the target hybrid optimization algorithm in the embodiment of the present application with the multi-objective optimization algorithm in the related art. The target hybrid optimization algorithm in the embodiment of the present application has a higher optimization effect for target workflow scheduling, and the algorithm performance is better, the algorithm convergence is stronger, and the diversity and uniformity of the Pareto solution distribution are better.

[0139] Figure 3 This is a schematic diagram of the structure of a workflow scheduling device provided in an embodiment of the present application. Figure 3 , the workflow scheduling device 30 may include:

[0140] An acquisition module 31 is used to acquire target attribute data corresponding to a target workflow;

[0141] A construction module 32 is used to construct a multi-objective optimization model corresponding to the target workflow according to the target attribute data; the multi-objective optimization model includes a hybrid optimization algorithm;

[0142] A determination module 33, used to determine a target solution set corresponding to a target workflow according to target attribute data and a multi-objective optimization model;

[0143] The decoding module 34 is used to decode the target solution set to obtain the target workflow scheduling mode corresponding to the target workflow.

[0144] In a possible implementation, the target attribute data includes execution task data, computing resource data, execution time data, execution cost data, and load balancing data.

[0145] In a possible implementation, the construction module 32 is specifically configured to:

[0146] According to the execution task data, determine the directed acyclic graph corresponding to the target workflow, and determine the entry task node and the exit task node corresponding to the directed acyclic graph;

[0147] Determine the target workflow scheduling model corresponding to the target workflow according to the computing resource data and the directed acyclic graph;

[0148] According to the execution time data, execution cost data, load balancing data and the target workflow scheduling model, a multi-objective optimization model corresponding to the target workflow scheduling model is constructed.

[0149] In a possible implementation manner, the determination module 33 is specifically configured to:

[0150] Initialize the multi-objective optimization model according to the execution task data corresponding to the target workflow;

[0151] According to the computing resource data, the directed acyclic graph corresponding to the target workflow and the mapping relationship between computing resources are encoded and processed to obtain the initial feasible solution corresponding to the multi-objective optimization model;

[0152] The initial feasible solution is calculated and processed by the target hybrid optimization algorithm in the multi-objective optimization model to obtain the target solution set corresponding to the target workflow.

[0153] In a possible implementation, the target hybrid optimization algorithm is a genetic bacterial foraging optimization algorithm; the genetic bacterial foraging optimization algorithm includes a genetic optimization algorithm and a bacterial foraging optimization algorithm; the determination module 33 is specifically used to:

[0154] According to the genetic optimization algorithm, the initial feasible solution is iterated to obtain the intermediate solution set;

[0155] When the intermediate solution set meets the target switching condition, the intermediate solution set is iteratively calculated according to the bacterial foraging optimization algorithm to obtain the target solution set corresponding to the target workflow.

[0156] In a possible implementation, the genetic optimization algorithm includes a selection operator based on a Pareto dominance relationship, an adaptive crossover operator, and an adaptive mutation operator;

[0157] The target switching conditions include that the minimum average evolution rate of the population and the individual fitness variance of the population meet the preset conditions, or the number of iterations of the genetic optimization algorithm reaches the maximum number of iterations;

[0158] The bacterial foraging optimization algorithm includes a tendency operation with an adaptive tendency step size based on a step size adjustment factor and a dispersal operation based on grid division.

[0159] The workflow scheduling device 30 provided in the embodiment of the present application can execute the technical solution shown in the above method embodiment, and its implementation principle and beneficial effects are similar, which will not be repeated here.

[0160] Figure 4 This is a schematic diagram of the structure of a workflow scheduling device provided in an embodiment of the present application. Figure 4The workflow scheduling device 40 may include: a memory 41 and a processor 42. Exemplarily, the memory 41 and the processor 42 are interconnected via a bus 43.

[0161] The memory 41 is used to store program instructions;

[0162] The processor 42 is used to execute the program instructions stored in the memory to implement the workflow scheduling method shown in the above embodiment.

[0163] Figure 4 The workflow scheduling device 40 shown can execute the technical solution shown in the above method embodiment, and its implementation principle and beneficial effects are similar, which will not be repeated here.

[0164] An embodiment of the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the above-mentioned workflow scheduling method.

[0165] The embodiment of the present application may also provide a computer program product, including a computer program, which, when executed by a processor, can implement the above-mentioned workflow scheduling method.

[0166] It should be noted that the processor mentioned in the embodiments of the present application may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0167] It should be understood that the memory mentioned in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM) and direct RAM bus random access memory (DR RAM). It should be noted that when the processor is a general purpose processor, DSP, ASIC, FPGA or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, the memory (storage module) is integrated in the processor. It should be noted that the memory described herein is intended to include but is not limited to these and any other suitable types of memory.

[0168] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0169] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processing unit of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processing unit of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0170] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0171] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0172] Regarding the various modules / units contained in the various devices and products described in the above embodiments, they may be software modules / units, or hardware modules / units, or they may be partially software modules / units and partially hardware modules / units. Each device and product may be applied to or integrated in a chip, a chip module or a terminal device. Exemplarily, for each device and product applied to or integrated in a chip, each module / chip contained therein may be implemented in the form of hardware such as circuits, or at least some modules / units may be implemented in the form of software programs, which run on a processor integrated inside the chip, and the remaining modules / units may be implemented in the form of hardware such as circuits.

[0173] In the present application, the term "include" and its variations may refer to non-restrictive inclusion; the term "or" and its variations may refer to "and / or". The terms "first", "second", etc. in the present application are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. In the present application, "plurality" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the previously associated objects are in an "or" relationship.

[0174] The above are only some embodiments of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A workflow scheduling method, characterized in that: include: Obtain target attribute data corresponding to the target workflow; Constructing a multi-objective optimization model corresponding to the target workflow according to the target attribute data; The multi-objective optimization model includes an objective hybrid optimization algorithm; Determining a target solution set corresponding to the target workflow according to the target attribute data and the multi-objective optimization model; The target solution set is decoded to obtain a target workflow scheduling method corresponding to the target workflow.

2. The method according to claim 1, characterized in that The target attribute data includes execution task data, computing resource data, execution time data, execution cost data and load balancing data.

3. The method according to claim 2, characterized in that The step of constructing a multi-objective optimization model corresponding to the target workflow according to the target attribute data includes: Determine, according to the execution task data, a directed acyclic graph corresponding to the target workflow, and determine an entry task node and an exit task node corresponding to the directed acyclic graph; Determining a target workflow scheduling model corresponding to the target workflow according to the computing resource data and the directed acyclic graph; A multi-objective optimization model corresponding to the target workflow scheduling model is constructed according to the execution time data, the execution cost data, the load balancing data and the target workflow scheduling model.

4. The method according to claim 2, characterized in that: Determining a target solution set corresponding to the target workflow according to the target attribute data and the multi-objective optimization model includes: Initializing the multi-objective optimization model according to the execution task data corresponding to the target workflow; According to the computing resource data, encoding processing is performed on the directed acyclic graph corresponding to the target workflow and the mapping relationship between computing resources to obtain an initial feasible solution corresponding to the multi-objective optimization model; The initial feasible solution is calculated and processed by the target hybrid optimization algorithm in the multi-objective optimization model to obtain a target solution set corresponding to the target workflow.

5. The method according to claim 4, characterized in that The target hybrid optimization algorithm is a genetic bacterial foraging optimization algorithm; the genetic bacterial foraging optimization algorithm includes a genetic optimization algorithm and a bacterial foraging optimization algorithm; the target hybrid optimization algorithm in the multi-objective optimization model is used to calculate the initial feasible solution to obtain the target solution set corresponding to the target workflow, including: According to the genetic optimization algorithm, the initial feasible solution is iterated to obtain an intermediate solution set; When the intermediate solution set meets the target switching condition, the intermediate solution set is iteratively calculated according to the bacterial foraging optimization algorithm to obtain the target solution set corresponding to the target workflow.

6. The method according to claim 5, characterized in that The genetic optimization algorithm includes a selection operator based on Pareto dominance relationship, an adaptive crossover operator and an adaptive mutation operator; The target switching condition includes that the minimum average evolution rate of the population and the individual fitness variance of the population meet the preset conditions, or the number of iterations of the genetic optimization algorithm reaches the maximum number of iterations; The bacterial foraging optimization algorithm includes a trending operation using an adaptive trending step size based on a step size adjustment factor and a dispersing operation based on grid division.

7. A workflow scheduling device, characterized in that: include: An acquisition module is used to acquire target attribute data corresponding to a target workflow; A construction module, used to construct a multi-objective optimization model corresponding to the target workflow according to the target attribute data; the multi-objective optimization model includes a hybrid optimization algorithm; A determination module, used to determine a target solution set corresponding to the target workflow according to the target attribute data and the multi-objective optimization model; The decoding module is used to decode the target solution set to obtain the target workflow scheduling mode corresponding to the target workflow.

8. A workflow scheduling device, characterized in that: include: Memory, processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory, so that the processor executes the workflow scheduling method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement the workflow scheduling method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the workflow scheduling method according to any one of claims 1 to 6 is implemented.

Images