CN116700928A

CN116700928A - Task scheduling information generation method, device, equipment and storage medium

Info

Publication number: CN116700928A
Application number: CN202310721431.6A
Authority: CN
Inventors: 曾胜泓
Original assignee: CCB Finetech Co Ltd
Current assignee: CCB Finetech Co Ltd
Priority date: 2023-06-16
Filing date: 2023-06-16
Publication date: 2023-09-05

Abstract

The disclosure provides a method, a device, equipment and a storage medium for generating task scheduling information, which can be applied to the technical field of computers and the technical field of finance. The method comprises the following steps: acquiring resource demand information of a task to be scheduled and resource use state information of a server cluster; generating a plurality of first candidate task scheduling information according to the resource demand information and the resource use state information; the method comprises the steps that a plurality of first candidate task scheduling information represents a plurality of resource allocation proportion information of a server cluster to which a task to be scheduled is allocated; carrying out adaptability evaluation on the plurality of first candidate task scheduling information to obtain a adaptability evaluation result; based on a genetic algorithm, processing a plurality of first candidate task scheduling information according to a fitness evaluation result to obtain a first candidate task scheduling information set; and optimizing the first candidate task scheduling information set based on an annealing algorithm to obtain target task scheduling information.

Description

Task scheduling information generation method, device, equipment and storage medium

Technical Field

The present disclosure relates to the field of computer technology and the field of financial technology, and in particular, to a method, apparatus, device, medium, and program product for generating task scheduling information.

Background

Task scheduling refers to automatically executing tasks based on preset time points, preset time intervals or preset execution times.

In the related art, when algorithms are adopted for computing power distribution, a single algorithm may have some side effects in computing power distribution, for example, a genetic algorithm may cause population convergence to a locally optimal solution based on genetic variation and fitness selection of individuals, a problem that a globally optimal solution cannot be achieved, abnormal task scheduling occurs, even execution of other task schedules is affected, and unacceptable results are generated in the whole system category.

Disclosure of Invention

In view of the foregoing, the present disclosure provides a task scheduling information generation method, apparatus, device, medium, and program product.

According to a first aspect of the present disclosure, there is provided a task scheduling information generating method, including: acquiring resource demand information of a task to be scheduled and resource use state information of a server cluster;

generating a plurality of first candidate task scheduling information according to the resource demand information and the resource use state information; the first candidate task scheduling information characterizes a plurality of resource allocation proportion information of the server cluster to which the task to be scheduled is allocated;

Performing fitness evaluation on the plurality of first candidate task scheduling information to obtain a fitness evaluation result;

based on a genetic algorithm, processing the plurality of first candidate task scheduling information according to the fitness evaluation result to obtain a first candidate task scheduling information set; and

and carrying out optimization processing on the first candidate task scheduling information set based on an annealing algorithm to obtain target task scheduling information.

According to an embodiment of the present disclosure, the performing the fitness evaluation on the plurality of first candidate task scheduling information to obtain a fitness evaluation result includes:

evaluating the predicted task processing duration of the plurality of first candidate task scheduling information to obtain a first evaluation result;

evaluating the predicted resource utilization rate of the plurality of first candidate task scheduling information to obtain a second evaluation result; and

and obtaining the fitness evaluation result according to the first evaluation result and the second evaluation result.

According to an embodiment of the present disclosure, the task to be scheduled includes N, where N is an integer greater than 1, and the evaluating the predicted task processing durations of the plurality of first candidate task scheduling information to obtain a first evaluation result includes:

Generating the total task processing time length of N tasks to be scheduled in each first candidate task scheduling information according to the resource allocation proportion information and the resource demand information aiming at each first candidate task scheduling information;

and generating the first evaluation result according to the total task processing time length of the N tasks to be scheduled.

According to an embodiment of the present disclosure, the server cluster includes M servers, where M is an integer greater than 1, and the evaluating the predicted resource utilization of the plurality of first candidate task scheduling information to obtain a second evaluation result includes:

generating resource utilization rate information of M servers according to the resource allocation proportion information aiming at each first candidate task scheduling information;

generating average resource utilization information according to the resource utilization information of the M servers; and

and generating the second evaluation result according to the average resource utilization information.

According to an embodiment of the present disclosure, the obtaining the fitness evaluation result according to the first evaluation result and the second evaluation result includes:

determining a first weight corresponding to the first evaluation result and a second weight corresponding to the second evaluation result; and

And obtaining the fitness evaluation result according to the first weight, the first evaluation result, the second evaluation result and the second weight.

According to an embodiment of the disclosure, the processing, based on the genetic algorithm, the plurality of first candidate task scheduling information according to the fitness evaluation result to obtain a first candidate task scheduling information set includes:

obtaining a selection probability interval corresponding to the plurality of first candidate task scheduling information according to the fitness evaluation result;

obtaining a plurality of second candidate task scheduling information from the plurality of first candidate task scheduling information based on the random number generated randomly; and

and obtaining the first candidate task scheduling information set according to the plurality of second candidate task scheduling information.

According to an embodiment of the present disclosure, the task scheduling information generating method further includes:

sequencing the plurality of first candidate task scheduling information according to the fitness evaluation result to obtain a sequencing result;

based on the sorting result, a plurality of third candidate task scheduling information is obtained from the plurality of first candidate task scheduling information; and

and obtaining the first candidate task scheduling information set according to the second candidate task scheduling information and the third candidate task scheduling information.

randomly grouping the plurality of first candidate task scheduling information to obtain a plurality of candidate task scheduling information groups, wherein each candidate task scheduling information group comprises at least two first candidate task scheduling information;

exchanging partial resource allocation proportion information in at least two first candidate task scheduling information in the candidate task scheduling information groups for each candidate task scheduling information group to obtain a plurality of fourth candidate task scheduling information; and

and obtaining the first candidate task scheduling information set according to the second candidate task scheduling information, the third candidate task scheduling information and the fourth candidate task scheduling information.

changing the resource allocation proportion of the first candidate task scheduling information according to a preset rule to obtain a fifth candidate task scheduling information;

the first candidate task scheduling information set is obtained according to the second candidate task scheduling information, the third candidate task scheduling information, the fourth candidate task scheduling information and the fifth candidate task scheduling information.

According to an embodiment of the present disclosure, the optimizing the first candidate task scheduling information set based on the annealing algorithm to obtain target task scheduling information includes:

taking any one of the first candidate task scheduling information in the first candidate task scheduling information set as an initial value;

adjusting the resource allocation proportion in the first candidate task scheduling information corresponding to the initial value to obtain a change value;

obtaining a probability of the change value being selected based on an objective function from the initial value and the change value;

when the probability is determined to be larger than a preset threshold value, changing first candidate task scheduling information corresponding to an initial value in the first candidate task scheduling information set into first candidate task scheduling information corresponding to the changed value to obtain a second candidate task scheduling information set; and

and sequencing the candidate task scheduling information in the second candidate task scheduling set according to the fitness evaluation result to obtain the target task scheduling information.

A second aspect of the present disclosure provides a task scheduling information generating apparatus, including:

The acquisition module is used for acquiring resource demand information of the task to be scheduled and resource use state information of the server cluster;

the first generation module is used for generating a plurality of first candidate task scheduling information according to the resource demand information and the resource use state information; the first candidate task scheduling information characterizes a plurality of resource allocation proportion information of the server cluster to which the task to be scheduled is allocated;

the evaluation module is used for carrying out fitness evaluation on the plurality of first candidate task scheduling information to obtain a fitness evaluation result;

the first processing module is used for processing the plurality of first candidate task scheduling information according to the adaptability evaluation result based on a genetic algorithm to obtain a first candidate task scheduling information set; and

and the second processing module is used for carrying out optimization processing on the first candidate task scheduling information set based on an annealing algorithm to obtain target task scheduling information.

According to an embodiment of the present disclosure, wherein the evaluation module comprises:

the duration evaluation sub-module is used for evaluating the predicted task processing duration of the plurality of first candidate task scheduling information to obtain a first evaluation result;

The utilization rate evaluation sub-module is used for evaluating the predicted resource utilization rate of the plurality of first candidate task scheduling information to obtain a second evaluation result; and

and the result generation sub-module is used for generating the adaptability evaluation result according to the first evaluation result and the second evaluation result.

According to the embodiment of the disclosure, the tasks to be scheduled include N, where N is an integer greater than 1.

The duration evaluation submodule comprises:

the total time length generation unit is used for generating the task processing total time length of N tasks to be scheduled in each first candidate task scheduling information according to the resource allocation proportion information and the resource demand information aiming at each first candidate task scheduling information;

and the first evaluation result generating unit is used for generating the first evaluation result according to the total task processing time length of the N tasks to be scheduled.

According to the embodiment of the disclosure, the server cluster includes M servers, where M is an integer greater than 1.

The utilization rate evaluation submodule includes:

a resource utilization information generating unit, configured to generate resource utilization information of M servers according to the resource allocation proportion information for each first candidate task scheduling information;

An average resource utilization information generating unit, configured to generate average resource utilization information according to the resource utilization information of the M servers; and

and a second evaluation result generating unit configured to generate the second evaluation result according to the average resource utilization information.

According to an embodiment of the present disclosure, wherein the result generation submodule includes:

a determining unit configured to determine a first weight corresponding to the first evaluation result and a second weight corresponding to the second evaluation result; and

and the fitness evaluation result generating unit is used for generating the fitness evaluation result according to the first weight, the first evaluation result, the second evaluation result and the second weight.

According to an embodiment of the disclosure, the first processing module includes:

the interval generation sub-module is used for generating selection probability intervals corresponding to the plurality of first candidate task scheduling information according to the adaptability evaluation result;

an information generation sub-module for generating a plurality of second candidate task schedule information from the plurality of first candidate task schedule information based on the randomly generated random number; and

and the information set generation sub-module is used for generating the first candidate task scheduling information set according to the plurality of second candidate task scheduling information.

According to an embodiment of the present disclosure, the task scheduling information generating apparatus further includes:

the sorting module is used for sorting the plurality of first candidate task scheduling information according to the fitness evaluation result to obtain a sorting result;

a second generation module, configured to generate a plurality of third candidate task schedule information from the plurality of first candidate task schedule information based on the sorting result; and

and a third generating module, configured to generate the first candidate task scheduling information set according to the plurality of second candidate task scheduling information and the plurality of third candidate task scheduling information.

the grouping module is used for randomly grouping the plurality of first candidate task scheduling information to obtain a plurality of candidate task scheduling information groups, wherein each candidate task scheduling information group comprises at least two first candidate task scheduling information;

the exchange module is used for exchanging part of resource allocation proportion information in at least two first candidate task scheduling information in the candidate task scheduling information groups for each candidate task scheduling information group to obtain a plurality of fourth candidate task scheduling information; and

And a fourth generating module, configured to generate the first candidate task scheduling information set according to the plurality of second candidate task scheduling information, the plurality of third candidate task scheduling information, and the plurality of fourth candidate task scheduling information.

the changing module is used for changing the resource allocation proportion of the plurality of first candidate task scheduling information according to a preset rule to obtain a plurality of fifth candidate task scheduling information;

and a fifth generating module, configured to generate the first candidate task scheduling information set according to the second candidate task scheduling information, the third candidate task scheduling information, the fourth candidate task scheduling information, and the fifth candidate task scheduling information.

According to an embodiment of the disclosure, the second processing module includes:

the selecting sub-module is used for selecting any one of the first candidate task scheduling information in the first candidate task scheduling information set as an initial value;

an adjustment sub-module, configured to adjust a resource allocation ratio in the first candidate task scheduling information corresponding to the initial value, to obtain a change value;

A probability generation sub-module for generating a probability that the change value is selected based on the initial value and the change value based on an objective function;

a changing sub-module, configured to change, when it is determined that the probability is greater than a predetermined threshold, first candidate task scheduling information corresponding to an initial value in the first candidate task scheduling information set to first candidate task scheduling information corresponding to the changed value, to obtain a second candidate task scheduling information set; and

and the sequencing sub-module is used for sequencing the candidate task scheduling information in the second candidate task scheduling set according to the adaptability evaluation result to obtain the target task scheduling information.

A third aspect of the present disclosure provides an electronic device, comprising: one or more processors; and a memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to perform the method.

A fourth aspect of the present disclosure also provides a computer-readable storage medium having stored thereon executable instructions that, when executed by a processor, cause the processor to perform the above-described method.

A fifth aspect of the present disclosure also provides a computer program product comprising a computer program which, when executed by a processor, implements the above method.

According to the task scheduling information generation method, the device, the equipment, the medium and the program product, resource demand information of a task to be scheduled and resource use state information of a server cluster are obtained; generating a plurality of first candidate task scheduling information according to the resource demand information and the resource use state information; carrying out adaptability evaluation on the plurality of first candidate task scheduling information to obtain a adaptability evaluation result; based on a genetic algorithm, processing a plurality of first candidate task scheduling information according to a fitness evaluation result to obtain a first candidate task scheduling information set; and optimizing the first candidate task scheduling information set based on an annealing algorithm to obtain target task scheduling information. Due to the combination of the genetic algorithm and the annealing algorithm, the individual population of the scheduling scheme is enlarged, the generation of the local optimal solution can be effectively reduced, the utilization rate of the whole resources is improved, and the flexible sharing of the resources and the higher success rate of task processing are realized.

Drawings

The foregoing and other objects, features and advantages of the disclosure will be more apparent from the following description of embodiments of the disclosure with reference to the accompanying drawings, in which:

FIG. 1 schematically illustrates an application scenario diagram of a task scheduling information generation method, apparatus, device, medium and program product according to an embodiment of the present disclosure;

FIG. 2 schematically illustrates a flow chart of a task scheduling information generation method according to an embodiment of the present disclosure;

FIG. 3 schematically illustrates a flow chart of a method of fitness evaluation of a plurality of first candidate task schedule information according to an embodiment of the disclosure;

FIG. 4 schematically illustrates a flow chart of a method of optimizing a first set of candidate task scheduling information according to an embodiment of the disclosure;

FIG. 5 schematically illustrates a flow diagram of another method of task scheduling information generation according to an embodiment of the disclosure;

fig. 6 schematically illustrates a system architecture diagram of a task scheduling information generation method according to an embodiment of the present disclosure;

fig. 7 schematically illustrates a block diagram of a task scheduling information generating apparatus according to an embodiment of the present disclosure; and

fig. 8 schematically illustrates a block diagram of an electronic device adapted to implement a task scheduling information generation method according to an embodiment of the present disclosure.

Detailed Description

Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. It should be understood that the description is only exemplary and is not intended to limit the scope of the present disclosure. In the following detailed description, for purposes of explanation, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the present disclosure. It may be evident, however, that one or more embodiments may be practiced without these specific details. In addition, in the following description, descriptions of well-known structures and techniques are omitted so as not to unnecessarily obscure the concepts of the present disclosure.

The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. The terms "comprises," "comprising," and/or the like, as used herein, specify the presence of stated features, steps, operations, and/or components, but do not preclude the presence or addition of one or more other features, steps, operations, or components.

All terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art unless otherwise defined. It should be noted that the terms used herein should be construed to have meanings consistent with the context of the present specification and should not be construed in an idealized or overly formal manner.

Where expressions like at least one of "A, B and C, etc. are used, the expressions should generally be interpreted in accordance with the meaning as commonly understood by those skilled in the art (e.g.," a system having at least one of A, B and C "shall include, but not be limited to, a system having a alone, B alone, C alone, a and B together, a and C together, B and C together, and/or A, B, C together, etc.).

In the technical scheme of the disclosure, the related data (such as including but not limited to personal information of a user) are collected, stored, used, processed, transmitted, provided, disclosed, applied and the like, all conform to the regulations of related laws and regulations, necessary security measures are adopted, and the public welcome is not violated.

In the related art, when an algorithm is adopted to perform calculation power distribution in a task scheduling process, a single algorithm may have some side effects in calculation power distribution, for example, a genetic algorithm may cause a population to converge on a local optimal solution based on genetic variation and fitness selection of an individual, and a problem that global optimal solution cannot be achieved, and abnormal task scheduling occurs may be caused. Based on the above, the embodiment of the disclosure provides a task scheduling information generation method, which can integrate a genetic algorithm and an annealing algorithm, expand individual population of a scheduling scheme, effectively reduce generation of local optimal solutions, finally improve utilization rate of overall resources, and realize flexible sharing of resources and higher success rate of task processing.

Embodiments of the present disclosure provide a task scheduling information generation method, apparatus, device, storage medium, and computer program product. The task scheduling information generation method comprises the following steps: acquiring resource demand information of a task to be scheduled and resource use state information of a server cluster; generating a plurality of first candidate task scheduling information according to the resource demand information and the resource use state information; the method comprises the steps that a plurality of first candidate task scheduling information represents a plurality of resource allocation proportion information of a server cluster to which a task to be scheduled is allocated; carrying out adaptability evaluation on the plurality of first candidate task scheduling information to obtain a adaptability evaluation result; based on a genetic algorithm, processing a plurality of first candidate task scheduling information according to a fitness evaluation result to obtain a first candidate task scheduling information set; and optimizing the first candidate task scheduling information set based on an annealing algorithm to obtain target task scheduling information.

Fig. 1 schematically illustrates an application scenario diagram of a task scheduling information generation method, apparatus, device, medium and program product according to an embodiment of the present disclosure.

As shown in fig. 1, an application scenario 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, and a third terminal device 103. The network 104 is a medium used to provide a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.

The user may interact with the server 105 through the network 104 using at least one of the first terminal device 101, the second terminal device 102, the third terminal device 103, to receive or send messages, etc. Various communication client applications, such as a shopping class application, a web browser application, a search class application, an instant messaging tool, a mailbox client, social platform software, etc. (by way of example only) may be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103.

The first terminal device 101, the second terminal device 102, the third terminal device 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smartphones, tablets, laptop and desktop computers, and the like.

The server 105 may be a server providing various services, such as a background management server (by way of example only) providing support for websites browsed by the user using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process the received data such as the user request, and feed back the processing result (e.g., the web page, information, or data obtained or generated according to the user request) to the terminal device.

It should be noted that, the task scheduling information generating method provided by the embodiments of the present disclosure may be generally executed by the server 105. Accordingly, the task scheduling information generating apparatus provided by the embodiments of the present disclosure may be generally provided in the server 105. The task scheduling information generation method provided by the embodiment of the present disclosure may also be performed by a server or a server cluster that is different from the server 105 and is capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and/or the server 105. Accordingly, the task scheduling information generating apparatus provided by the embodiments of the present disclosure may also be provided in a server or a server cluster that is different from the server 105 and is capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and/or the server 105.

It should be understood that the number of terminal devices, networks and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.

The task scheduling information generation method of the disclosed embodiment will be described in detail below with reference to fig. 2 to 6 based on the scenario described in fig. 1.

Fig. 2 schematically illustrates a flowchart of a task scheduling information generation method according to an embodiment of the present disclosure.

As shown in fig. 2, the task scheduling information generation method of this embodiment includes operations S210 to S250.

In operation S210, resource requirement information of a task to be scheduled and resource usage state information of a server cluster are acquired.

According to embodiments of the present disclosure, the resource demand information may include basic information of the server and demand information of the task to be scheduled for the computing power resource. The basic information of the servers can be the number of the servers and the computing power range of the servers, and the basic information comprises parameters such as CPU model numbers, memory sizes, disk capacities and the like. The parameter information can be acquired through a server management tool, an API interface or other modes, the information is recorded in the system, and a corresponding server basic information record is established for subsequent calculation power distribution. The information of the demand of the task to be regulated on the computing power resource can be that the system obtains the demand of the task to be regulated on the computing power resource according to each task in the task queue information, wherein the demand comprises CPU computing capacity and memory space which are occupied by each task. The task queue may be a centrally managed queue containing tasks to be executed. The demand information of the computing power resources can also be determined according to the type, the priority, the resource demand and other information of the tasks.

According to embodiments of the present disclosure, the resource usage status information may include server usage status information and remaining computing power calculation information. The server usage status information may be a usage status of the server periodically monitored by the system, including CPU usage, memory usage, disk space, and the like. These may be implemented by monitoring tools, agents, or other means. The current resource occupation condition of each server can be known through monitoring. The remaining computing power computing information may be computing the remaining computing power of each server according to the basic information of the server cluster and the task currently running, and these may be computed by comprehensively considering the indexes such as CPU utilization rate, memory utilization condition, and the like. The specific calculation method may be determined according to the system requirements and the calculation force distribution strategy.

In some possible embodiments, the calculation information of the remaining calculation force may include: CPU remaining computing power information, memory remaining computing power information and disk remaining computing power information. The specific method can be calculated according to the task type and the resource consumption model. For example, in calculating the remaining calculation power of the CPU, the remaining calculation power of the CPU is calculated according to the CPU model of the server and the CPU occupation amount of the current task, and factors such as the CPU core number, the main frequency and the parallel processing capacity of the server can be considered; in the process of calculating the remaining memory computing power, the remaining memory computing power is calculated according to the memory size of the server and the memory occupation amount of the current task, and the total memory capacity of the server and the currently used memory amount can be considered; in the calculation of the remaining computing power of the disk, the remaining computing power of the disk is calculated according to the disk capacity of the server and the disk occupation amount of the current task, and the total disk capacity of the server and the currently used disk amount can be considered.

In operation S220, a plurality of first candidate task scheduling information is generated according to the resource demand information and the resource usage state information.

According to an embodiment of the present disclosure, the first candidate task schedule information may be candidate task schedule information generated by random sampling. The first candidate task schedule information may also be referred to as initial candidate task schedule information. Wherein random sampling can be performed by using a uniform distribution method or a normal distribution method. It should be noted that the sum of the first candidate task schedule information for each server should be ensured to meet the total first candidate task schedule information requirement. And if the requirement is not met, the first candidate task scheduling information is adjusted or regenerated. For example, it is checked whether each first candidate task schedule information is within a reasonable range and whether the total calculation force meets the requirements of the task demand. And if the first candidate task scheduling information does not meet the constraint condition, regenerating.

In one possible embodiment, the above operations are repeated until a sufficient amount of first candidate task schedule information is generated as the initial candidate task schedule information set. In general, the size of the first set of candidate task scheduling information depends on the size and complexity of the task to be scheduled, and a larger first set of candidate task scheduling information is generally suggested to increase the exploration capacity of the solution space.

For example: there are 3 servers in the server cluster, server a, server B and server C, respectively, each with a computing power in the range of 0% to 100%. A set of first candidate task schedule information is generated using a genetic algorithm, each first candidate task schedule information representing a computational power distribution scheme for a server. The process of generating an individual is as follows: for each solution, the calculated force allocation proportion for each server is randomly generated. For example, the calculation force distribution ratio of the random generation server a is 50%, the calculation force distribution ratio of the server B is 30%, and the calculation force distribution ratio of the server C is 20%. These calculation force distribution ratios were combined into one scheme, expressed as [50%,30%,20% ].

In one possible embodiment, the resource allocation proportion information may be adjusted according to the type, priority, and resource requirements of the task, e.g., a computationally intensive task may require a large amount of CPU computing resources, while a storage intensive task may require more disk capacity, so the system needs to determine its requirements for different resources according to the type of task; among the priorities of tasks, high priority tasks typically need to be completed faster, and thus may require allocation of more computing power resources; some tasks have strict requirements on the completion time, e.g., real-time tasks or urgent tasks need to be completed within a specified time frame, so the system needs to consider the time requirements of the task and ensure that sufficient computing power resources are allocated to meet the time constraints of the task; in addition, the task has difference on the resource consumption degree, some tasks may consume higher resources such as CPU, memory, disk and the like, and other tasks may consume lower resources, so the system needs to consider the consumption condition of the tasks on various resources and reasonably allocate according to the condition of available resources.

In operation S230, the fitness evaluation is performed on the plurality of first candidate task scheduling information, so as to obtain a fitness evaluation result.

According to an embodiment of the present disclosure, the fitness evaluation may calculate the fitness of each first candidate task scheduling information using a fitness function. The fitness function should be defined according to the characteristics and the targets of the problem, such as task completion time, resource utilization rate, and the like.

Fig. 3 schematically illustrates a flowchart of performing fitness evaluation on a plurality of first candidate task schedule information to obtain a fitness evaluation result according to an embodiment of the present disclosure.

As shown in fig. 3, the first candidate task scheduling information fitness evaluation method of this embodiment includes operations S231 to S234.

In operation S231, an objective function is defined. An objective function is determined based on the task scheduling objectives. For example, the task completion time may be minimized as an objective function, or the indexes such as the task completion time and the resource utilization may be comprehensively considered, and combined into one comprehensive index as an objective function.

In operation S232, each first candidate task scheduling information is evaluated. For each first candidate task schedule information, simulations or simulations are performed to evaluate its performance. And simulating the execution process of the task according to the first candidate task scheduling information, and calculating corresponding indexes.

In one possible embodiment, the corresponding metrics may include task completion time, resource utilization, and the like.

In operation S233, a fitness value is calculated. And converting the index value obtained by evaluation into a fitness value according to the objective function and the corresponding index. The specific calculation mode can be selected according to the nature of the index and the characteristics of the problem.

For example: the task completion time and the resource utilization rate are taken as two indexes of fitness evaluation, and are given different weights.

Calculation of task completion time (TFT):

there may be three tasks currently pending, which need to be done on servers A, B and C, respectively.

For each task, the time it takes to complete on the corresponding server is calculated based on the type, priority and time requirements of the task.

For example: task 1 takes 2 hours on server a, task 2 takes 3 hours on server B, and task 3 takes 1 hour on server C.

The total task completion time was 2+3+1=6 hours.

Calculation of Resource Utilization (RU):

for example: the total computing power of servers A, B and C is 100, 200, and 150, respectively.

Under the current calculation force distribution scheme, a server A distributes 50 calculation forces, a server B distributes 80 calculation forces, and a server C distributes 60 calculation forces.

Calculating the utilization rate of resources, namely the ratio of the total calculation power of the allocated tasks to the total calculation power of the server:

resource utilization of server a: 50/100=0.5

Resource utilization of server B: 80/200=0.4

Resource utilization of server C: 60/150=0.4

Average resource utilization is (0.5+0.4+0.4)/3=0.43

According to the above example, an fitness function may be defined, the specific operation being as shown in formula (1):

fitness function = w1 x (1/TFT) +w2 x RU (1)

Wherein w1 and w2 are weight coefficients for adjusting the importance of task completion time and resource utilization. And calculating the fitness function to obtain the fitness value of each first candidate task scheduling information (calculation power distribution scheme) for guiding the evolution process of the genetic algorithm.

In operation S234, a fitness value is assigned. And distributing the calculated fitness value to corresponding first candidate task scheduling information so as to perform selection, crossing and mutation operations in the subsequent genetic algorithm evolution process.

Through the operation, the adaptability evaluation is carried out on the plurality of first candidate task scheduling information, and finally the adaptability evaluation result is obtained.

In operation S240, a plurality of first candidate task scheduling information is processed according to the fitness evaluation result based on the genetic algorithm, to obtain a first candidate task scheduling information set.

According to embodiments of the present disclosure, processing the plurality of first candidate task schedule information may be using a selection, crossover and mutation operation of a genetic algorithm to evolve the plurality of first candidate task schedule information. For example, the selecting operation may be to select the first candidate task scheduling information with higher fitness as the parent first candidate task scheduling information, and then perform the interleaving and mutation operation to generate the child first candidate task scheduling information. The selecting, crossing and mutating operations are repeated until a next generation first candidate task scheduling information set is generated.

In a possible embodiment, the Selection is chosen, i.e. the server unit with the higher fitness is chosen as parent first candidate task scheduling information based on the result of the fitness evaluation. Common selection methods include roulette selection, tournament selection, and the like. By selecting the operation, the server unit with higher adaptability can be reserved, and the chance of reproduction thereof can be improved.

Roulette wheel bet selection (Roulette Wheel Selection): in roulette selection, the probability of selection of the first candidate task schedule information is proportional to its fitness value. The method comprises the following specific steps: calculating the sum of fitness values of each first candidate task scheduling information in the first candidate task scheduling information set, and taking the sum as the total scale of the wheel disc; for each first candidate task scheduling information, calculating the proportion of the adaptability value of each first candidate task scheduling information to the total scale to obtain the selection probability; a random number is generated, and the selected first candidate task scheduling information is determined according to the selection probability.

Tournament selection (Tournament Selection): in the tournament selection, a set of first candidate task schedule information (referred to as a tournament) is randomly selected, and then the first candidate task schedule information having the highest fitness is selected from the tournament as parent first candidate task schedule information. The method comprises the following specific steps: randomly selecting a number (tournament scale) of first candidate task scheduling information as participants of the tournament; the first candidate task scheduling information with the highest fitness is selected from the tournament participants as parent first candidate task scheduling information.

In one possible embodiment, cross over (crosslever) is an operation in a genetic algorithm for generating child first candidate task schedule information. In server computing power distribution, a cross operation may be applied to a computing power distribution scheme of a server.

The method specifically comprises the following operations:

and randomly selecting parent first candidate task scheduling information. And randomly selecting two pieces of first candidate task scheduling information from the parent first candidate task scheduling information obtained in the previous step of selecting operation as crossed parents. These parent first candidate task schedule information represent different server computing power allocation schemes.

And (5) performing cross operation. In the crossover operation, we select a certain crossover point from the two parent first candidate task schedule information and crossover their calculation power distribution schemes. The specific operation is as follows: first, a cross point is selected, which may be a position or a range, and the cross point may be selected randomly or according to a certain strategy; then, before or after the intersection, the computing power distribution schemes of the two parent first candidate task scheduling information are intersected, for example, the computing power distribution part of a certain server is exchanged, or the task is transferred from one server to another server;

according to an embodiment of the present disclosure, two child first candidate task schedule information will be generated that inherit some of the characteristics of parent first candidate task schedule information, in combination with different computing power allocation schemes. Through the cross operation, diversified child first candidate task scheduling information can be generated to explore different first candidate task scheduling information. This facilitates better exploration of the genetic algorithm in the search space and increases the probability of finding a globally optimal solution.

In one possible embodiment, mutation (Mutation) is an operation in a genetic algorithm to introduce new genetic information and increase diversity of populations. In the server computing power distribution, the mutation operation may be applied to a computing power distribution scheme of the server.

In a possible embodiment, the evolution process is repeated, i.e. the selection, crossing and mutation operations are repeated, new child first candidate task schedule information is generated, and the population (first candidate task schedule information set) is updated. Typically, the evolution process will proceed for multiple generations until a predetermined termination condition is reached, such as a maximum number of iterations is reached or a solution is found that meets the requirements.

In one possible embodiment, updating the population is a key step in the genetic algorithm for forming the next generation population. In task scheduling, the update population may be an update of the first set of candidate task scheduling information.

The specific operation comprises the following steps: the generated child first candidate task scheduling information generates a plurality of new child first candidate task scheduling information through selection, crossing and mutation operations, wherein the first candidate task scheduling information represents improved first candidate task scheduling information obtained through genetic algorithm operation; original parent first candidate task schedule information, which represents initial first candidate task schedule information; and merging the population, and merging the generated child first candidate task scheduling information with the original parent first candidate task scheduling information to form a new first candidate task scheduling information set. Thus, the next generation population is composed of the first candidate task schedule information subjected to the selection, crossover and mutation operations.

In operation S250, optimization processing is performed on the first candidate task scheduling information set based on the annealing algorithm, so as to obtain target task scheduling information.

According to the embodiment of the disclosure, the optimization processing mainly includes performing local search operation on part of the first candidate task scheduling information by using an optimization algorithm, that is, selecting a certain amount of first candidate task scheduling information, and performing an iterative search process on a corresponding server computing power distribution scheme.

Fig. 4 schematically illustrates a flowchart of optimizing a first candidate task scheduling information set to obtain target task scheduling information according to an embodiment of the disclosure.

As shown in fig. 4, the method for optimizing the first candidate task scheduling information set according to this embodiment includes operations S251 to S255. The specific operation is as follows.

In operation S251, one first candidate task scheduling information is selected as a current solution;

in operation S252, minor changes are made based on the current solution, which may include fine tuning the computing power allocation of a certain server, or reassigning tasks to different servers;

in operation S253, the server utilization of the new solution is calculated, which may be calculated by simulating the execution process of the task, taking into consideration the type, priority, and time requirements of the task, and the calculation power allocation of the server.

In operation S254, the utilization of the new solution is compared with the utilization of the current solution. If the utilization rate of the new solution is better than that of the current solution, accepting the new solution as the current solution, and continuing the next optimization; if the new solution has a worse utilization value than the current solution, the system accepts the solution with larger difference with a certain probability. Initially, the probability of accepting a solution with a larger variance is higher; as the iteration progresses, the probability of accepting a solution with a larger difference gradually decreases, so that the algorithm tends to converge to a more optimal solution. The probability calculation formula is specifically shown as formula (2):

P_accept ＝ exp((C_current - C_new) / N) (2)

wherein, the initial value of N is 100, iterate once N-1.

C_current represents the server residual utilization of the current solution: e.g. 15%, we are rounded to 15.

C_new represents the server remaining utilization when new solution: e.g. 10%, we are rounded to 10.

While the probability acceptance value (preset value) may be 90.

At the beginning of the iteration, p_accept=exp ((10-15)/100) ≡exp (-0.05) ≡0.951)

At this time, the probability is larger than a preset value, and the scheme is acceptable.

With iteration, N will gradually decrease.

When n=50, c_current=7, the new solution is c_new=12. Substituting these values into the formula again calculates the probability:

P_accept＝exp((7-12)/50)≈exp(-0.1)≈0.905

When the iteration proceeds further, when n=10, c_current=5, c_new=8. Calculating the probability:

P_accept＝exp((5-8)/10)≈exp(-0.3)≈0.741

the probability is reduced to about 74.1% and less than the preset probability value, and the scheme is not acceptable.

It can be seen that as the number of iterations increases, the probability of accepting a solution with a larger difference gradually decreases. Initially, the algorithm is more receptive to solutions with large differences, facilitating exploration in the solution space. With the reduction of the N value, the algorithm gradually converges to a better solution, and the probability of acceptance of solutions with larger differences is reduced, so that the optimal solution is more stably searched.

In operation S255, operations S252 to S254 are repeated until a termination condition is reached, for example, the maximum number of iterations is reached or a certain degree of convergence is reached.

According to the embodiment of the disclosure, the resource requirement information of the task to be scheduled and the resource use state information of the server cluster are obtained; generating a plurality of first candidate task scheduling information according to the resource demand information and the resource use state information; carrying out adaptability evaluation on the plurality of first candidate task scheduling information to obtain a adaptability evaluation result; based on a genetic algorithm, processing a plurality of first candidate task scheduling information according to a fitness evaluation result to obtain a first candidate task scheduling information set; and optimizing the first candidate task scheduling information set based on an annealing algorithm to obtain target task scheduling information. Due to the combination of the genetic algorithm and the annealing algorithm, the individual population of the scheduling scheme is enlarged, the generation of the local optimal solution can be effectively reduced, the utilization rate of the whole resources is finally improved, and the flexible sharing of the resources and the higher success rate of task processing are realized.

Fig. 5 schematically illustrates a flowchart of another task scheduling information generation method according to an embodiment of the present disclosure.

According to an embodiment of the present disclosure, performing fitness evaluation on a plurality of first candidate task scheduling information to obtain a fitness evaluation result, and specifically as shown in fig. 5, the task scheduling information generating method may include the following operations S510 to S530.

In operation S510, the predicted task processing durations of the plurality of first candidate task scheduling information are evaluated, and a first evaluation result is obtained.

In operation S520, the predicted resource utilization of the plurality of first candidate task scheduling information is evaluated, and a second evaluation result is obtained.

In operation S530, an fitness evaluation result is obtained according to the first evaluation result and the second evaluation result.

According to an embodiment of the present disclosure, the fitness evaluation index may include: predicted task processing duration and predicted resource utilization. In specific operations, mapping can be performed according to the predicted task processing duration, the smaller duration corresponds to a higher fitness value, and the first evaluation result is a calculation result of the fitness of the predicted task processing duration. Meanwhile, mapping can be performed according to the predicted resource utilization rate, a higher resource utilization rate corresponds to a higher fitness value, and the second evaluation result is a predicted resource utilization rate fitness calculation result. It should be noted that, if there are comprehensive indexes, the fitness value of each index may be comprehensively calculated by using methods such as weighted summation, weighted average, and the like.

According to an embodiment of the present disclosure, a task to be scheduled includes N, where N is an integer greater than 1, and the evaluating the predicted task processing durations of the plurality of first candidate task scheduling information to obtain a first evaluation result includes: aiming at each first candidate task scheduling information, generating the total task processing time length of N tasks to be scheduled in each first candidate task scheduling information according to the resource allocation proportion information and the resource demand information; and obtaining a first evaluation result according to the total task processing time length of the N tasks to be scheduled.

In a possible embodiment, the resource allocation proportion information may be a calculated power allocation proportion in a certain first candidate task scheduling information, for example, the resource allocation proportion information of the server a is 60. The resource requirement information may be resource requirement information in certain first candidate task scheduling information, for example, the server a allocates 2 scheduling tasks, each scheduling task requiring 30 computing power resource requirement information. The total task processing duration of the N tasks to be scheduled may be total duration required by the N tasks to be scheduled in a certain first candidate task scheduling information, for example, the resource allocation proportion information of the server in a certain first candidate task scheduling information is 60, there are 2 scheduling tasks, and the calculation power requirement of each scheduling task is 30, so the total resource requirement information is 30×2=60, the total required task processing duration is 1 hour according to the resource allocation proportion information of the server, the resource allocation proportion information satisfies the resource requirement information, and the total task processing duration is the first evaluation result.

According to an embodiment of the present disclosure, a server cluster includes M servers, where M is an integer greater than 1, and the evaluating the predicted resource utilization of the plurality of first candidate task scheduling information to obtain a second evaluation result includes: aiming at each first candidate task scheduling information, obtaining resource utilization rate information of M servers according to the resource allocation proportion information; generating average resource utilization information according to the resource utilization information of the M servers; and generating a second evaluation result according to the average resource utilization information.

In a possible embodiment, the resource utilization information may be a resource utilization value of M servers calculated according to the resource allocation proportion information, for example, the resource utilization of 3 servers in a certain first candidate task scheduling information is 90%, 80% and 70%, respectively. The average resource utilization information may be calculated by performing average calculation on the resource utilization information of M servers, for example, the resource utilization of 3 servers in a certain first candidate task scheduling information is 90%, 80% and 70%, and the average resource utilization information of M servers is 80% obtained by calculation, where the average resource utilization information is the second evaluation result.

According to an embodiment of the present disclosure, obtaining an fitness evaluation result according to a first evaluation result and a second evaluation result includes: determining a first weight corresponding to the first evaluation result and a second weight corresponding to the second evaluation result; and obtaining an adaptability evaluation result according to the first weight, the first evaluation result, the second evaluation result and the second weight.

In a possible embodiment, the fitness evaluation result may be calculated based on the first evaluation result and the second evaluation result and the respective corresponding weight values.

According to the embodiment of the disclosure, according to the fitness evaluation result, an evaluation result value can be allocated to each matched first candidate task scheduling information so as to facilitate selection, crossing and mutation operations in the subsequent genetic algorithm evolution process.

According to an embodiment of the present disclosure, based on a genetic algorithm, processing a plurality of first candidate task scheduling information according to a fitness evaluation result to obtain a first candidate task scheduling information set, including: obtaining a selection probability interval corresponding to the plurality of first candidate task scheduling information according to the fitness evaluation result; based on the random number generated randomly, a plurality of second candidate task scheduling information is obtained from the plurality of first candidate task scheduling information; and obtaining a first candidate task scheduling information set according to the plurality of second candidate task scheduling information.

In a possible embodiment, the selection probability interval may be determined according to the ranking of the first candidate task scheduling information, and the selection probability may be calculated by using a linear scale or an exponential scale, so that the first candidate task scheduling information with higher fitness has higher selection probability. The random number may be a value between 0 and 1 that is randomly generated. The second candidate task scheduling information may be selected according to a random number, that is, a plurality of first candidate task scheduling information is randomly selected, so as to obtain a plurality of second candidate task scheduling information.

According to an embodiment of the present disclosure, the task scheduling information generating method further includes: sequencing the plurality of first candidate task scheduling information according to the fitness evaluation result to obtain a sequencing result; based on the sorting result, a plurality of third candidate task scheduling information is obtained from the plurality of first candidate task scheduling information; and obtaining a first candidate task scheduling information set according to the plurality of second candidate task scheduling information and the plurality of third candidate task scheduling information.

In a possible embodiment, the ranking may be that the plurality of first candidate task scheduling information is ranked according to the fitness evaluation result thereof, and ranking is performed from the first candidate task scheduling information with the highest fitness to the first candidate task scheduling information with the lowest fitness. The third candidate task scheduling information may be selected according to the selection probability by sequentially traversing the first candidate task scheduling information from the ordered first candidate task scheduling information list, and when the cumulative probability exceeds the random number, the corresponding first candidate task scheduling information is selected as the third candidate task scheduling information, which may also be referred to as parent first candidate task scheduling information. It should be noted that, a plurality of third candidate task schedule information is generated as needed, and the selection operation is repeated until a sufficient number of third candidate task schedule information is generated.

According to an embodiment of the present disclosure, the task scheduling information generating method further includes: randomly grouping a plurality of first candidate task scheduling information to obtain a plurality of candidate task scheduling information groups, wherein each candidate task scheduling information group comprises at least two first candidate task scheduling information; exchanging partial resource allocation proportion information in at least two first candidate task scheduling information in the candidate task scheduling information groups aiming at each candidate task scheduling information group to obtain a plurality of fourth candidate task scheduling information; and obtaining a first candidate task scheduling information set according to the plurality of second candidate task scheduling information, the plurality of third candidate task scheduling information and the plurality of fourth candidate task scheduling information.

In a possible embodiment, the exchanging of part of the resource allocation proportion information in at least two first candidate task schedule information in the candidate task schedule information group may be to utilize interleaving, to partly exchange the resource allocation proportion information of a certain server, or to transfer the task schedule from one server to another server. The fourth candidate task scheduling information may be at least two child first candidate task scheduling information generated by the exchanging operation, where the child first candidate task scheduling information inherits part of the characteristics of the parent child first candidate task scheduling information, and combines different computing power allocation schemes.

According to the embodiment of the disclosure, through the operation, diversified child first candidate task scheduling information can be generated to explore different first candidate task scheduling information, so that a genetic algorithm can be better explored in a search space, and the probability of finding a globally optimal solution is improved.

According to an embodiment of the present disclosure, the task scheduling information generating method further includes: changing the resource allocation proportion of the first candidate task scheduling information according to a preset rule to obtain a plurality of fifth candidate task scheduling information; and obtaining a first candidate task scheduling information set according to the plurality of second candidate task scheduling information, the plurality of third candidate task scheduling information, the plurality of fourth candidate task scheduling information and the plurality of fifth candidate task scheduling information.

In a possible embodiment, the changing the resource allocation proportion according to the predetermined rule may be an operation of changing the resource allocation proportion of the server by using a mutation operation, and the specific operation includes: and selecting one first candidate task scheduling information from the plurality of child first candidate task scheduling information obtained by the cross operation as a variant object. The first candidate task scheduling information characterizes new first candidate task scheduling information obtained through cross operation; modifying the selected first candidate task scheduling information to introduce new first candidate task scheduling information, wherein the specific steps comprise: randomly selecting one or more first candidate task scheduling information in a computing power distribution scheme of the first candidate task scheduling information; and according to the strategy of the mutation operation, mutating the selected first candidate task scheduling information. For example, it may be to increase or decrease the power allocation proportion of the server.

According to the embodiment of the disclosure, through the mutation operation, the diversity of the first candidate task scheduling information set is increased, and premature sinking into a local optimal solution is avoided. By introducing new first candidate task scheduling information, the mutation operation enables the first candidate task scheduling information to be more diversified, and therefore more search space is provided.

According to an embodiment of the present disclosure, based on an annealing algorithm, performing optimization processing on a first candidate task scheduling information set to obtain target task scheduling information, including: taking any one of the first candidate task scheduling information in the first candidate task scheduling information set as an initial value; adjusting the resource allocation proportion in the first candidate task scheduling information corresponding to the initial value to obtain a change value; based on the objective function, obtaining the probability of the change value being selected according to the initial value and the change value; under the condition that the probability is larger than a preset threshold value, changing first candidate task scheduling information corresponding to an initial value in the first candidate task scheduling information set into first candidate task scheduling information corresponding to a changed value to obtain a second candidate task scheduling information set; and sequencing the candidate task scheduling information in the second candidate task scheduling set according to the fitness evaluation result to obtain target task scheduling information.

In one possible embodiment, the initial value is based on each server, and the current resource allocation proportion value is calculated, and the resource allocation proportion value can be realized by accumulating the computing power resources required by the allocated tasks on the server. The change value may be a new resource allocation ratio generated randomly by adjusting (including fine-tuning the computing power allocation of a certain server or reallocating the tasks to a different server) the resource allocation ratio of the first candidate task scheduling information corresponding to the initial value. The objective function may be a calculated force usage rate for each server, the calculated force usage rate representing a degree to which the server is currently utilized, the value being between 0 and 1, and being calculated by dividing the calculated force usage amount for each server by the total calculated force for the server. The probability of the change value being selected can be obtained by weighted average of the computational power utilization of all servers to obtain the server utilization of the whole system. Similarly, the probability of the initial value being selected may also be calculated. It should be noted that, depending on the priority of the task, the task type, or the task time requirement, different weights may be given to the computational power usage of each server, for example, the computational power usage of the server where the high priority task is located may have a greater weight.

In a possible embodiment, the probability that the change value is selected is greater than the probability that the initial value is selected, and the change value is taken as the current solution (the first candidate task scheduling information corresponding to the change value); if the probability of the change value being selected is smaller than the probability of the initial value being selected, a solution with a larger difference is accepted with a certain probability. Initially, the probability of accepting a solution with a larger variance is higher; as the iteration progresses, the probability of accepting a solution with a larger difference gradually decreases, so that the algorithm tends to converge to a more optimal solution. Repeating the above operation, continuing to optimize until reaching the termination condition, such as reaching the maximum iteration number or reaching a certain convergence degree, and finally obtaining the second candidate task scheduling information set.

In a possible embodiment, the iteration of the algorithm is stopped when the termination condition is reached, according to a preset termination condition. And finally outputting a computing power distribution scheme corresponding to the second candidate task scheduling information with the highest fitness in the second candidate task scheduling information set as an approximate optimal solution for solving the problem of computing power distribution of the server.

In a possible embodiment, the selection of the optimal solution may be performed by elite retention policy, that is, when combining the new candidate task scheduling information obtained by searching through the optimization algorithm and the initial candidate task scheduling information set, ensuring that the optimal individual in the initial candidate task scheduling information set is also included in the new candidate task scheduling information set. This ensures that at least one individual in the new candidate task scheduling information set has an optimal fitness value.

The elite retention strategy comprises the following specific steps:

firstly, calculating the fitness value of each candidate task scheduling information in an initial candidate task scheduling information set; sorting the initial candidate task scheduling information according to the fitness value, namely arranging from the optimal to the worst; copying the candidate task scheduling information (usually one or more) with the optimal fitness value into a new candidate task scheduling information set, so as to ensure that the candidate task scheduling information is reserved; and merging the candidate task scheduling information obtained by searching through the optimization algorithm into a new candidate task scheduling information set, filling the remaining gaps to ensure that the size of the new candidate task scheduling information set is the same as that of the initial candidate task scheduling information set so as to keep the stability of the scale of the candidate task scheduling information set.

According to an embodiment of the present disclosure, by the above-described operations, the acquisition of the optimal server power allocation scheme is not obtained by reducing the number in the population (task scheduling information set) to do the elimination. The schemes in the population are updated step by step to be optimal solutions through iterative loop population algorithm and local search operation. Finally, the iteration times, the iteration time and the fitness value reach a threshold value or no obvious improvement is caused to be used as an ending condition after a period of time, and the population is in a final form.

as shown in fig. 6, first, initializing a first candidate task scheduling information set, performing fitness evaluation on each first candidate task scheduling information in the first candidate task scheduling information set, and performing selection, intersection and mutation operations on the first candidate task scheduling information set based on a genetic algorithm until a predetermined termination condition is reached to form a new first candidate task scheduling information set; then selecting a new first candidate task scheduling information through local search operation, forming a change value (new calculation power distribution scheme) through fine adjustment parameters, calculating the selected probability of the change value, determining to add the scheme into a new candidate task scheduling information set based on a preset threshold value by comparing the selected probability of the new calculation power distribution scheme and the old calculation power distribution scheme, and updating the population when the maximum iteration number or a certain convergence degree is reached; and finally, determining an optimal solution based on the related strategy, and outputting the optimal solution.

Based on the task scheduling information generation method, the disclosure also provides a task scheduling information generation device. The device will be described in detail below in connection with fig. 7.

Fig. 7 schematically shows a block diagram of a task scheduling information generating apparatus according to an embodiment of the present disclosure.

As shown in fig. 7, the task scheduling information generating apparatus 700 of this embodiment includes an acquisition module 710, a first generation module 720, an evaluation module 730, a first processing module 740, and a second processing module 750.

An obtaining module 710, configured to obtain resource requirement information of a task to be scheduled and resource usage state information of a server cluster; in an embodiment, the obtaining module 710 may be configured to perform the operation S210 described above, which is not described herein.

A first generating module 720, configured to generate a plurality of first candidate task scheduling information according to the resource requirement information and the resource usage status information; the method comprises the steps that a plurality of first candidate task scheduling information represents a plurality of resource allocation proportion information of a server cluster to which a task to be scheduled is allocated; in an embodiment, the first generating module 720 may be used to perform the operation S220 described above, which is not described herein.

The evaluation module 730 is configured to perform fitness evaluation on the plurality of first candidate task scheduling information, to obtain a fitness evaluation result; in an embodiment, the evaluation module 730 may be configured to perform the operation S230 described above, which is not described herein.

The first processing module 740 is configured to process, based on a genetic algorithm, the plurality of first candidate task scheduling information according to the fitness evaluation result, to obtain a first candidate task scheduling information set; in an embodiment, the first processing module 740 may be configured to perform the operation S240 described above, which is not described herein.

The second processing module 750 is configured to perform optimization processing on the first candidate task scheduling information set based on an annealing algorithm, so as to obtain target task scheduling information; in an embodiment, the second processing module 750 may be used to perform the operation S250 described above, which is not described herein.

According to an embodiment of the present disclosure, wherein the evaluation module comprises: the system comprises a duration evaluation sub-module, a utilization rate evaluation sub-module and a result generation sub-module.

And the duration evaluation sub-module is used for evaluating the predicted task processing duration of the plurality of first candidate task scheduling information to obtain a first evaluation result.

And the utilization rate evaluation sub-module is used for evaluating the predicted resource utilization rates of the plurality of first candidate task scheduling information to obtain a second evaluation result.

And the result generation sub-module is used for generating an adaptability evaluation result according to the first evaluation result and the second evaluation result.

According to the embodiment of the disclosure, the tasks to be scheduled comprise N, wherein N is an integer greater than 1.

The duration evaluation submodule comprises: a total duration generating unit and a first evaluation result generating unit.

The total time length generation unit is used for generating the total time length of task processing of N tasks to be scheduled in each first candidate task scheduling information according to the resource allocation proportion information and the resource demand information aiming at each first candidate task scheduling information.

And the first evaluation result generation unit is used for generating a first evaluation result according to the total task processing time length of the N tasks to be scheduled.

According to the embodiment of the disclosure, the server cluster comprises M servers, and M is an integer greater than 1.

The utilization rate evaluation submodule includes: the device comprises a resource utilization rate information generating unit, an average resource utilization rate information generating unit and a second evaluation result generating unit.

And the resource utilization rate information generation unit is used for generating resource utilization rate information of M servers according to the resource allocation proportion information for each first candidate task scheduling information.

And the average resource utilization information generating unit is used for generating average resource utilization information according to the resource utilization information of the M servers.

And the second evaluation result generating unit is used for generating a second evaluation result according to the average resource utilization rate information.

According to an embodiment of the present disclosure, wherein the result generation submodule includes: a determining unit and an adaptability estimation result generating unit.

And the determining unit is used for determining a first weight corresponding to the first evaluation result and a second weight corresponding to the second evaluation result.

And the fitness evaluation result generation unit is used for generating a fitness evaluation result according to the first weight, the first evaluation result, the second evaluation result and the second weight.

According to an embodiment of the disclosure, the first processing module includes: the system comprises a section generation sub-module, an information generation sub-module and an information set generation sub-module.

And the interval generation sub-module is used for generating selection probability intervals corresponding to the plurality of first candidate task scheduling information according to the adaptability evaluation result.

And the information generation sub-module is used for generating a plurality of second candidate task scheduling information from the plurality of first candidate task scheduling information based on the randomly generated random number.

And the information set generation sub-module is used for generating a first candidate task scheduling information set according to the plurality of second candidate task scheduling information.

According to an embodiment of the present disclosure, the task scheduling information generating apparatus further includes: the device comprises a sequencing module, a second generation module and a third generation module.

And the sequencing module is used for sequencing the plurality of first candidate task scheduling information according to the fitness evaluation result to obtain a sequencing result.

And the second generation module is used for generating a plurality of third candidate task scheduling information from the plurality of first candidate task scheduling information based on the sorting result.

And the third generation module is used for generating a first candidate task scheduling information set according to the plurality of second candidate task scheduling information and the plurality of third candidate task scheduling information.

According to an embodiment of the present disclosure, the task scheduling information generating apparatus further includes: a grouping module, a switching module and a fourth generating module.

The grouping module is used for randomly grouping the plurality of first candidate task scheduling information to obtain a plurality of candidate task scheduling information groups, wherein each candidate task scheduling information group comprises at least two first candidate task scheduling information.

The exchanging module is used for exchanging part of resource allocation proportion information in at least two first candidate task scheduling information in the candidate task scheduling information groups aiming at each candidate task scheduling information group to obtain a plurality of fourth candidate task scheduling information.

And the fourth generation module is used for generating a first candidate task scheduling information set according to the second candidate task scheduling information, the third candidate task scheduling information and the fourth candidate task scheduling information.

According to an embodiment of the present disclosure, the task scheduling information generating apparatus further includes: a change module and a fifth generation module.

The changing module is used for changing the resource allocation proportion of the first candidate task scheduling information according to a preset rule to obtain a plurality of fifth candidate task scheduling information;

and the fifth generation module is used for generating a first candidate task scheduling information set according to the second candidate task scheduling information, the third candidate task scheduling information, the fourth candidate task scheduling information and the fifth candidate task scheduling information.

According to an embodiment of the disclosure, the second processing module includes: the device comprises a selecting sub-module, an adjusting sub-module, a probability generating sub-module, a changing sub-module and a sorting sub-module.

And the selecting sub-module is used for selecting any one of the first candidate task scheduling information in the first candidate task scheduling information set as an initial value.

And the adjustment sub-module is used for adjusting the resource allocation proportion in the first candidate task scheduling information corresponding to the initial value to obtain a change value.

And the probability generation sub-module is used for generating the probability of the change value being selected according to the initial value and the change value based on the objective function.

And the change sub-module is used for changing the first candidate task scheduling information corresponding to the initial value in the first candidate task scheduling information set into the first candidate task scheduling information corresponding to the changed value under the condition that the determination probability is larger than the preset threshold value, so as to obtain a second candidate task scheduling information set.

And the sequencing sub-module is used for sequencing the candidate task scheduling information in the second candidate task scheduling set according to the fitness evaluation result to obtain target task scheduling information.

According to embodiments of the present disclosure, any of the acquisition module 710, the first generation module 720, the evaluation module 730, the first processing module 740, and the second processing module 750 may be combined in one module to be implemented, or any of the modules may be split into a plurality of modules. Alternatively, at least some of the functionality of one or more of the modules may be combined with at least some of the functionality of other modules and implemented in one module. According to embodiments of the present disclosure, at least one of the acquisition module 710, the first generation module 720, the evaluation module 730, the first processing module 740, and the second processing module 750 may be implemented at least in part as hardware circuitry, such as a Field Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a system on a chip, a system on a substrate, a system on a package, an Application Specific Integrated Circuit (ASIC), or may be implemented in hardware or firmware in any other reasonable way of integrating or packaging the circuitry, or in any one of or a suitable combination of any of the three implementations of software, hardware, and firmware. Alternatively, at least one of the acquisition module 710, the first generation module 720, the evaluation module 730, the first processing module 740, and the second processing module 750 may be at least partially implemented as computer program modules, which when executed, may perform the respective functions.

As shown in fig. 8, an electronic device 800 according to an embodiment of the present disclosure includes a processor 801 that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) 802 or a program loaded from a storage section 808 into a Random Access Memory (RAM) 803. The processor 801 may include, for example, a general purpose microprocessor (e.g., a CPU), an instruction set processor and/or an associated chipset and/or a special purpose microprocessor (e.g., an Application Specific Integrated Circuit (ASIC)), or the like. The processor 801 may also include on-board memory for caching purposes. The processor 801 may include a single processing unit or multiple processing units for performing the different actions of the method flows according to embodiments of the disclosure.

In the RAM 803, various programs and data required for the operation of the electronic device 800 are stored. The processor 801, the ROM802, and the RAM 803 are connected to each other by a bus 804. The processor 801 performs various operations of the method flow according to the embodiments of the present disclosure by executing programs in the ROM802 and/or the RAM 803. Note that the program may be stored in one or more memories other than the ROM802 and the RAM 803. The processor 801 may also perform various operations of the method flows according to embodiments of the present disclosure by executing programs stored in the one or more memories.

According to an embodiment of the present disclosure, the electronic device 800 may also include an input/output (I/O) interface 805, the input/output (I/O) interface 805 also being connected to the bus 804. The electronic device 800 may also include one or more of the following components connected to the I/O interface 805: an input portion 806 including a keyboard, mouse, etc.; an output portion 807 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and a speaker; a storage section 808 including a hard disk or the like; and a communication section 809 including a network interface card such as a LAN card, a modem, or the like. The communication section 809 performs communication processing via a network such as the internet. The drive 810 is also connected to the I/O interface 805 as needed. A removable medium 811 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 810 as needed so that a computer program read out therefrom is mounted into the storage section 808 as needed.

The present disclosure also provides a computer-readable storage medium that may be embodied in the apparatus/device/system described in the above embodiments; or may exist alone without being assembled into the apparatus/device/system. The computer-readable storage medium carries one or more programs which, when executed, implement methods in accordance with embodiments of the present disclosure.

According to embodiments of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, which may include, for example, but is not limited to: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this disclosure, a computer-readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. For example, according to embodiments of the present disclosure, the computer-readable storage medium may include ROM 802 and/or RAM 803 and/or one or more memories other than ROM 802 and RAM 803 described above.

Embodiments of the present disclosure also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowcharts. The program code, when executed in a computer system, causes the computer system to implement the task scheduling information generation method provided by the embodiments of the present disclosure.

The above-described functions defined in the system/apparatus of the embodiments of the present disclosure are performed when the computer program is executed by the processor 801. The systems, apparatus, modules, units, etc. described above may be implemented by computer program modules according to embodiments of the disclosure.

In one embodiment, the computer program may be based on a tangible storage medium such as an optical storage device, a magnetic storage device, or the like. In another embodiment, the computer program may also be transmitted, distributed, and downloaded and installed in the form of a signal on a network medium, and/or from a removable medium 811 via a communication portion 809. The computer program may include program code that may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the foregoing.

In such an embodiment, the computer program may be downloaded and installed from a network via the communication section 809, and/or installed from the removable media 811. The above-described functions defined in the system of the embodiments of the present disclosure are performed when the computer program is executed by the processor 801. The systems, devices, apparatus, modules, units, etc. described above may be implemented by computer program modules according to embodiments of the disclosure.

According to embodiments of the present disclosure, program code for performing computer programs provided by embodiments of the present disclosure may be written in any combination of one or more programming languages, and in particular, such computer programs may be implemented in high-level procedural and/or object-oriented programming languages, and/or assembly/machine languages. Programming languages include, but are not limited to, such as Java, c++, python, "C" or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, partly on a remote computing device, or entirely on the remote computing device or server. In the case of remote computing devices, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., connected via the Internet using an Internet service provider).

The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

Those skilled in the art will appreciate that the features recited in the various embodiments of the disclosure and/or in the claims may be provided in a variety of combinations and/or combinations, even if such combinations or combinations are not explicitly recited in the disclosure. In particular, the features recited in the various embodiments of the present disclosure and/or the claims may be variously combined and/or combined without departing from the spirit and teachings of the present disclosure. All such combinations and/or combinations fall within the scope of the present disclosure.

The embodiments of the present disclosure are described above. However, these examples are for illustrative purposes only and are not intended to limit the scope of the present disclosure. Although the embodiments are described above separately, this does not mean that the measures in the embodiments cannot be used advantageously in combination. The scope of the disclosure is defined by the appended claims and equivalents thereof. Various alternatives and modifications can be made by those skilled in the art without departing from the scope of the disclosure, and such alternatives and modifications are intended to fall within the scope of the disclosure.

Claims

1. A task scheduling information generation method includes:

acquiring resource demand information of a task to be scheduled and resource use state information of a server cluster;

Generating a plurality of first candidate task scheduling information according to the resource demand information and the resource use state information; the first candidate task scheduling information characterizes a plurality of resource allocation proportion information of a server cluster to which the task to be scheduled is allocated;

2. The method of claim 1, wherein the performing fitness evaluation on the plurality of first candidate task scheduling information to obtain a fitness evaluation result includes:

And generating the fitness evaluation result according to the first evaluation result and the second evaluation result.

3. The method of claim 2, wherein the tasks to be scheduled include N, where N is an integer greater than 1, and the evaluating the predicted task processing durations of the plurality of first candidate task scheduling information to obtain a first evaluation result includes:

aiming at each first candidate task scheduling information, generating the total task processing time length of N tasks to be scheduled in each first candidate task scheduling information according to the resource allocation proportion information and the resource demand information;

and obtaining the first evaluation result according to the total task processing time length of the N tasks to be scheduled.

4. The method of claim 2, wherein the server cluster includes M servers, M is an integer greater than 1, and the evaluating the predicted resource utilization of the plurality of first candidate task scheduling information to obtain a second evaluation result includes:

5. The method of claim 2, wherein the generating the fitness evaluation result from the first evaluation result and the second evaluation result comprises:

and generating the fitness evaluation result according to the first weight, the first evaluation result, the second evaluation result and the second weight.

6. The method of claim 1, wherein the processing the plurality of first candidate task scheduling information based on the genetic algorithm according to the fitness evaluation result to obtain a first candidate task scheduling information set includes:

based on the random number generated randomly, a plurality of second candidate task scheduling information is obtained from the plurality of first candidate task scheduling information; and

7. The method of claim 6, further comprising:

sorting the plurality of first candidate task scheduling information according to the fitness evaluation result to obtain a sorting result;

and obtaining the first candidate task scheduling information set according to the plurality of second candidate task scheduling information and the plurality of third candidate task scheduling information.

8. The method of claim 6, further comprising:

exchanging partial resource allocation proportion information in at least two first candidate task scheduling information in each candidate task scheduling information group to obtain a plurality of fourth candidate task scheduling information; and

and generating the first candidate task scheduling information set according to the second candidate task scheduling information, the third candidate task scheduling information and the fourth candidate task scheduling information.

9. The method of claim 6, further comprising:

the resource allocation proportion of the first candidate task scheduling information is changed according to a preset rule, so that a plurality of fifth candidate task scheduling information is obtained;

and generating the first candidate task scheduling information set according to the second candidate task scheduling information, the third candidate task scheduling information, the fourth candidate task scheduling information and the fifth candidate task scheduling information.

10. The method of claim 1, wherein the optimizing the first candidate task scheduling information set based on the annealing algorithm to obtain target task scheduling information includes:

selecting any one of the first candidate task scheduling information in the first candidate task scheduling information set as an initial value;

generating a probability that the change value is selected based on an objective function according to the initial value and the change value;

under the condition that the probability is larger than a preset threshold value, changing first candidate task scheduling information corresponding to an initial value in the first candidate task scheduling information set into first candidate task scheduling information corresponding to the changed value to obtain a second candidate task scheduling information set; and

11. A task scheduling information generating apparatus comprising:

the first generation module is used for generating a plurality of first candidate task scheduling information according to the resource demand information and the resource use state information; the first candidate task scheduling information characterizes a plurality of resource allocation proportion information of a server cluster to which the task to be scheduled is allocated;

the first processing module is used for processing the plurality of first candidate task scheduling information according to the fitness evaluation result based on a genetic algorithm to obtain a first candidate task scheduling information set; and

12. An electronic device, comprising:

one or more processors;

storage means for storing one or more programs,

wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to perform the method of any of claims 1-10.

13. A computer readable storage medium having stored thereon executable instructions which, when executed by a processor, cause the processor to perform the method according to any of claims 1 to 10.

14. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1 to 10.