CN112379996A - Cloud computing resource scheduling optimization method based on dynamic target strategy and application - Google Patents

Cloud computing resource scheduling optimization method based on dynamic target strategy and application Download PDF

Info

Publication number
CN112379996A
CN112379996A CN202011095662.3A CN202011095662A CN112379996A CN 112379996 A CN112379996 A CN 112379996A CN 202011095662 A CN202011095662 A CN 202011095662A CN 112379996 A CN112379996 A CN 112379996A
Authority
CN
China
Prior art keywords
chromosome
resource scheduling
task
optimal
adaptive value
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202011095662.3A
Other languages
Chinese (zh)
Inventor
詹志辉
王子佳
陈宗淦
张军
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
South China University of Technology SCUT
Original Assignee
South China University of Technology SCUT
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by South China University of Technology SCUT filed Critical South China University of Technology SCUT
Priority to CN202011095662.3A priority Critical patent/CN112379996A/en
Publication of CN112379996A publication Critical patent/CN112379996A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
    • G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/04—Constraint-based CAD

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Computation (AREA)
  • Biomedical Technology (AREA)
  • Computer Hardware Design (AREA)
  • Physiology (AREA)
  • Genetics & Genomics (AREA)
  • Artificial Intelligence (AREA)
  • Geometry (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a cloud computing resource scheduling optimization method based on a dynamic objective strategy and application thereof, wherein the method comprises the following steps: initializing a population; evaluating an adaptive value by adopting the reciprocal of the task flow execution cost; constructing a dynamic target strategy, namely, obtaining no feasible scheduling scheme under the condition that the minimization cost based on the deadline constraint is taken as an optimization target, taking the task completion time as the optimization target until the feasible scheduling scheme is found, and then taking the minimization cost based on the deadline constraint as the optimization target again; selecting chromosome individuals for entering the next generation through roulette; carrying out chromosome crossing and mutation; constructing an optimal individual retention strategy; and when a feasible solution cannot be found after the feasible solution is found out and the specified algebra is executed, outputting the solution of the generated optimal chromosome. The invention can obtain a scheduling scheme with lower cost under the same deadline constraint and a feasible scheduling scheme under the smaller deadline constraint, and has stronger applicability.

Description

Cloud computing resource scheduling optimization method based on dynamic target strategy and application
Technical Field
The invention relates to the technical field of cloud computing resource scheduling, in particular to a cloud computing resource scheduling optimization method based on a dynamic objective strategy and application thereof.
Background
The National Institute of Standards and Technology (NIST) definition of cloud computing is: cloud computing is a pay-per-use model that provides available, convenient, on-demand network access into a configurable shared pool of computing resources (resources including networks, servers, storage, applications and services) that can be provisioned quickly, with little administrative effort, or interaction with service providers.
When the 'big data age' comes, the problems faced by cloud computing generally have a relatively large data volume, so that the scheduling of resources is of great importance, which is related to the cost and efficiency of cloud computing. Cloud computing resource scheduling has proven to be an NP-hard problem with many limitations, and it is almost impossible to obtain an optimal solution within polynomial time, and only a solution close to the optimal solution can be found. With the development of intelligent computing methods, extensive attention has been drawn to research and exploration on solving the cloud computing resource scheduling problem by using the intelligent computing method. Although the cloud resource scheduling problem has been widely researched, the Quality of Service (QoS) of users cannot be well satisfied, and the characteristics of cloud computing resources, namely elasticity and heterogeneity, are not well considered. On the other hand, many researchers have focused on minimizing the runtime of a task, a measure of the general effectiveness, but rarely take into account the cost of task performance. While in any commercial activity, cost is absolutely one of the non-negligible factors. Pursuing the fastest completion task often requires a greater investment, resulting in a reduction in overall efficiency. It is desirable to balance the relationship between cost and time to maximize economic efficiency.
The IEEE TRANSACTIONS ON CLOUD COMPUTING by Rodriguez and Buyya in 2014 proposed a model based ON a cutoff time to minimize the running cost and proposed a Particle Swarm Optimization (PSO) to solve this problem. However, in the process of learning a certain dimension of the particles to the optimal particle in the particle swarm optimization, the learning tendency of the resource number does not necessarily lead to a better result, and the convergence is fast but is easy to early converge.
Disclosure of Invention
In order to overcome the defects and shortcomings in the prior art, the invention provides a cloud computing resource scheduling optimization method based on a dynamic target strategy, which can obtain a scheduling scheme with lower cost under the constraint of the same deadline and can still obtain a feasible scheduling scheme under the constraint of smaller deadline.
The invention further provides a cloud computing resource scheduling optimization system based on the dynamic objective strategy.
A third object of the present invention is to provide a storage medium.
It is a fourth object of the invention to provide a computing device.
In order to achieve the purpose, the invention adopts the following technical scheme:
a cloud computing resource scheduling optimization method based on a dynamic objective strategy comprises the following steps:
initializing a population: randomly generating a resource scheduling sequence for each chromosome in the population;
and (3) evaluating an adaptive value: evaluating an adaptive value by adopting the reciprocal of the task flow execution cost;
dynamic target policy: if the feasible solution meeting the final deadline requirement does not exist during the first-generation initialization, the task flow execution time is taken as an optimization target, the adaptive value of the chromosome is evaluated by the task flow execution time in a reciprocal manner, the chromosome with smaller TET and higher adaptive value is selected in each iteration until the solution obtained by the chromosome appearing in the population meets the final deadline requirement, and the adaptive value is evaluated by the reciprocal of the task flow execution cost again;
selecting an operator: selecting chromosome individuals entering the next generation by roulette according to the fitness value of the chromosomes;
crossover and mutation of chromosomes: each chromosome is provided with a cross probability and a variation probability, a random number is generated for each resource scheduling sequence, and if the random number is smaller than the cross probability, the resource scheduling sequence is selected as one of the male parents of the cross;
generating a random number for each task of each resource scheduling sequence, wherein if the random number is less than the variation probability, the resource number occupied by the task is varied;
optimal individual retention strategy: in each generation, if the adaptive value of a chromosome in the current generation is higher than that of the current optimal chromosome, taking the chromosome in the current generation as a new optimal chromosome; otherwise, replacing the chromosome with the minimum fitness value in the current generation with the current optimal chromosome individual;
and when a feasible solution cannot be found after the feasible solution is found out and the specified algebra is executed, outputting the solution of the generated optimal chromosome.
As a preferred technical solution, the randomly generating a resource scheduling sequence for each chromosome in the population includes:
the corresponding resources and tasks are represented by the numbers of the resources and the tasks, and the ith gene of each chromosome represents the task tiResource being executed, for chromosome x, xiJ denotes task tiIn resource rjIs executed.
As a preferred technical solution, the chromosome individuals entering the next generation are selected by roulette, and the probability of the resource scheduling sequence being selected is:
Figure BDA0002723651920000031
fitnessi＝1/TECi
wherein TEC represents the task flow execution cost, fitnessiIndicates the fitness value, and i indicates the ith position of the chromosomal gene.
In the step of crossing and mutating chromosomes, each time 2 chromosomes are selected, a random number N ∈ [1, N-1], N ∈ [ Z ], is generated, and resource numbers of the first N tasks of the two resource scheduling sequences are correspondingly exchanged.
In order to achieve the second object, the present invention adopts the following technical solutions:
a cloud computing resource scheduling optimization system based on a dynamic objective strategy comprises: the system comprises a population initialization module, an adaptive value evaluation module, a dynamic target strategy construction module, a selection operator module, a chromosome crossing and mutation module, an optimal individual retention strategy construction module and an output module;
the population initialization module is used for initializing a population and randomly generating a resource scheduling sequence for each chromosome in the population;
the adaptive value evaluation module is used for evaluating an adaptive value by adopting the reciprocal of the task flow execution cost;
the dynamic target strategy construction module is used for constructing a dynamic target strategy, and specifically comprises the following steps: if the feasible solution meeting the final deadline requirement does not exist during the first-generation initialization, the task flow execution time is taken as an optimization target, the adaptive value of the chromosome is evaluated by the task flow execution time in a reciprocal manner, the chromosome with smaller TET and higher adaptive value is selected in each iteration until the solution obtained by the chromosome appearing in the population meets the final deadline requirement, and the adaptive value is evaluated by the reciprocal of the task flow execution cost again;
the selection operator module is used for selecting chromosome individuals entering the next generation through roulette according to the adaptive value of the chromosomes;
the chromosome crossing and mutation module is used for completing the crossing and mutation of chromosomes, each chromosome is provided with a crossing probability and a mutation probability, a random number is generated for each resource scheduling sequence, and if the random number is smaller than the crossing probability, the resource scheduling sequence is selected as one of the male parents of the crossing;
generating a random number for each task of each resource scheduling sequence, wherein if the random number is less than the variation probability, the resource number occupied by the task is varied;
the optimal individual retention strategy construction module is used for constructing an optimal individual retention strategy, and specifically comprises the following steps: in each generation, if the adaptive value of a chromosome in the current generation is higher than that of the current optimal chromosome, taking the chromosome in the current generation as a new optimal chromosome; otherwise, replacing the chromosome with the minimum fitness value in the current generation with the current optimal chromosome individual;
the output module is used for outputting an optimal solution, and outputting the generated optimal chromosome solution when a feasible solution cannot be found after a specified algebra is found.
In order to achieve the third object, the present invention adopts the following technical solutions:
a storage medium stores a program, and when executed by a processor, the program implements the above dynamic objective policy-based cloud computing resource scheduling optimization method.
In order to achieve the fourth object, the present invention adopts the following technical means:
a computing device comprises a processor and a memory for storing a processor executable program, and when the processor executes the program stored in the memory, the cloud computing resource scheduling optimization method based on the dynamic objective strategy is realized.
Compared with the prior art, the invention has the following advantages and beneficial effects:
(1) in the process of learning a certain dimension of particles in the existing particle swarm algorithm to the optimal particle, the tendency learning of resource numbers is not necessarily good and can lead to better results, and convergence is fast but premature convergence is easy.
(2) Because the genetic algorithm is not suitable for the optimization target with the constraint condition, the invention provides a dynamic target strategy, when a feasible scheduling scheme is not obtained (namely the deadline is exceeded) under the condition that the minimization cost based on the deadline constraint is taken as the optimization target, the task completion time is taken as the optimization target until the algorithm finds the feasible scheduling scheme, and then the minimization cost based on the deadline constraint is taken as the optimization target again, so that the applicability of the algorithm is stronger, and a feasible solution can still be found under the constraint of smaller deadline.
Drawings
FIG. 1 is a schematic diagram of an example cloud computing task flow according to the present invention;
FIG. 2 is a schematic diagram of an example runtime and data transfer time matrix according to the present invention;
FIG. 3 is a schematic diagram of a cloud computing task flow scheduling scheme according to the present invention;
FIG. 4 is a schematic diagram of a crossover in the genetic algorithm of the present invention;
FIG. 5 is a schematic representation of variations in the genetic algorithm of the present invention;
fig. 6 is a schematic flow diagram of a cloud computing resource scheduling optimization method based on a dynamic objective policy according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Examples
A user has a series of tasks to be executed, and the cloud computing platform provides a series of running resources (virtual machines) which can be rented by the user as required, so how to allocate the tasks to the corresponding resources to run forms a resource scheduling problem. As shown in fig. 1, it can be seen that there are many tasks in a task group, and there is a topology between these tasks, i.e. there are parent-child relationships between the tasks. The value on each edge represents the transfer time of the data, namely when the parent task and the child task do not run on the same resource, the parent task needs to transfer the data required by the running of some child tasks to the child tasks. The time required for each task to execute on each resource is also known. For programming implementation, the embodiment adopts numbers to represent tasks and resources, and defines two matrixes of exetime and transit time, wherein the matrix represents the task tiIn resource rjRun time of (i)][j]Representing a task tiTransmitting data to tjWhen i is j or tiAnd tjTransfer time [ i ] when there is no direct dependency, i.e. there are no directly connected edges in the topology][j]0. As shown in fig. 2, an example of two matrices, exetime and transit time, is given, and as shown in fig. 3, a representation of the progress of a task flow scheduling scheme over time is shown.
In the present embodiment, the objective model to be optimized is to minimize the execution cost based on the deadline, and its mathematical expressions are shown as (1) and (2), where TEC and TET represent the cost of task completion and the time required for task completion, respectively.
Minimize TEC (1)
TET<deadline (2)
The TEC and TET calculation methods are shown by (3) and (4), where r is given to each resourcejDefine its lease start time
Figure BDA0002723651920000071
And rental end time
Figure BDA0002723651920000072
Figure BDA0002723651920000073
Representing a task tiEnd time of execution; TET represents that the time of the task completed last in all the tasks of the resource is the time for completing the whole task flow; the cost is calculated by adding the price of each resource times the time it has been leased.
Figure BDA0002723651920000074
Figure BDA0002723651920000075
Last definitionA scheduling scheme for scheduling pos [ | T | ]]Denotes, pos [ i]The value of (b) represents the task tiRun on resource rpos[i]The above. Thus, TEC and TET can be calculated.
For each task tiFrom the pos sequence, the resource r on which it runs can be obtainedpos[i]。tiStart time of operation of
Figure BDA0002723651920000076
Is determined by two factors. If there is no parent, i.e. predecessor, then once resource r is availablepos[i]Is available then tiOperation can begin and this value corresponds to resource rpos[i]Lease end time LET ofrpos[i]. If t isiHaving one or more parent tasks, except for waiting for resource rpos[i]Available, yet can begin execution by waiting for all of its parent tasks to finish execution, this value corresponding to resource rpos[i]And the maximum value of the rental end time of (c) and the execution end time of all parent tasks. And the running time of this task
Figure BDA0002723651920000077
Is composed of a running time exetime [ i ]][rpos[i]]And data transfer time [ i ]][child(i)]And adding to obtain the final product. It should be noted that in practical applications, if the subtasks and t are usediThe data does not need the time of data transfer when the data runs on the same resource. And task tiEnd of run time of
Figure BDA0002723651920000078
Is its start time
Figure BDA0002723651920000079
Plus run time of tasks
Figure BDA00027236519200000710
If t isiResource r in which to runpos[i]Has not been selected by other tasks, then resource rpos[i]LST (r)rpos[i]Is the task tiStart time of operation of
Figure BDA00027236519200000711
At the same time, r is also required to bepos[i]To the set R of leased resources. The lease end time LET of the resource is then updated. Resource rpos[i]LET (L)rpos[i]Is task tiThe end time of the run. From the data calculated above, TET and TEC can be calculated from formula (1) and formula (2) in the model definition.
As shown in fig. 4, the present embodiment provides a cloud computing resource scheduling optimization method based on a dynamic objective policy, including the following steps:
(1) initializing a population
Randomly initializing each chromosome in the population to generate a resource scheduling sequence; according to the population definition of the genetic algorithm, each chromosome in the population corresponds to a solution of the problem, so each chromosome corresponds to each scheduling scheme, i.e., a scheduling sequence. In this embodiment, the numbers of resources and tasks are used to represent the corresponding resources and tasks, and the ith gene of each chromosome represents task tiThe resource in which this is performed, specifically, this embodiment encodes chromosomes with integers, for chromosome x, xiJ denotes task tiIn resource rjIs performed where i e [0, | T | -1]；
(2) Adaptive value evaluation
Evaluation formula of fitness value of each chromosome is as shown in (5), since the higher the fitness value in the genetic algorithm is, the better the chromosome is and the goal is to minimize the Cost, the reciprocal of the task flow Execution Cost TEC (TEC) is used as the fitness value of the chromosome. Due to the programming requirement, if the deadline of the solution obtained by the chromosome is not satisfactory, i.e. the task flow Execution Time (TET) of the solution exceeds the deadline, the TEC of the chromosome is defined as inf, a large number:
fitnessi＝1/TECi (5)
(3) dynamic object policy
If the feasible solution meeting the demand of deadline does not exist during the first generation initialization, the execution time of the task flow, namely TET is used as an optimization target, the adaptive value of the chromosome is evaluated by using the reciprocal of TET, the chromosome with smaller TET and higher adaptive value is selected in each iteration of the algorithm until the solution obtained by one chromosome in the population can meet the demand of deadline, namely the TET of the chromosome is smaller than the deadline, and the adaptive value is evaluated again by using the reciprocal of TEC, namely the formula (5);
(4) selection operator
Selecting chromosome individuals entering the next generation by roulette according to the fitness value of the chromosomes; the probability that the resource scheduling sequence i is selected is shown in formula (6);
Figure BDA0002723651920000091
(5) crossover and mutation of chromosomes
Each chromosome has a crossover probability PcAnd a mutation probability Pm(ii) a As shown in FIG. 5, for each chromosome, i.e., each resource scheduling sequence, a random number r ∈ [0, 1] is generated]If r is<PcSelecting the sequence as one of the parents of the cross; each time 2 chromosomes are selected, a random number N e [1, N-1] is generated]N belongs to Z, and correspondingly exchanges the resource numbers of the first n tasks of the two sequences;
as shown in FIG. 6, for each chromosome, i.e., each task of each resource scheduling sequence, a random number r e [0, 1] is generated]If r is<PmIf the number of the resource occupied by the task is changed, xm＝rand(0,N-1)；
(6) Optimal individual retention strategy
In each generation, if the adaptive value of the chromosome in the current generation is higher than the optimal chromosome, the chromosome is used as the optimal chromosome, otherwise, the worst chromosome in the current generation, namely the chromosome with the minimum adaptive value is replaced by the optimal chromosome individual, and the optimal chromosome individual is stored by one variable in the embodiment; the optimal solution generated in the operation process of the algorithm can be ensured not to be discarded by reserving the optimal individual, so that the algorithm is accelerated to find the optimal scheduling scheme;
(7) end conditions
And when a feasible solution is found and then executed to a specified algebra (4000 generations) or the feasible solution cannot be found within 100000 generations, outputting the optimal solution generated by the algorithm and ending the whole process.
Aiming at the problem that the genetic algorithm is not suitable for the optimization target with constraint conditions, the embodiment provides a dynamic target strategy; when a feasible scheduling scheme is not obtained under the condition that the minimization cost based on the deadline constraint is taken as an optimization target, the task completion time is taken as the optimization target, the scheduling scheme with less task completion time is selected each time, the minimization cost based on the deadline constraint is taken as the optimization target again after the algorithm finds the feasible scheduling scheme, the scheduling scheme with less task scheduling cost is selected each time, and the operation is circulated continuously until the program operation is finished; this strategy makes the algorithm more adaptive and still finds a feasible solution under the constraint of a smaller deadline.
Example 2
The embodiment provides a cloud computing resource scheduling optimization system based on a dynamic objective policy, which includes: the system comprises a population initialization module, an adaptive value evaluation module, a dynamic target strategy construction module, a selection operator module, a chromosome crossing and mutation module, an optimal individual retention strategy construction module and an output module;
in this embodiment, the population initialization module is configured to initialize a population, and randomly generate a resource scheduling sequence for each chromosome in the population;
in this embodiment, the adaptive value evaluation module is configured to evaluate the adaptive value by using a reciprocal of the task flow execution cost;
in this embodiment, the dynamic target policy constructing module is configured to construct a dynamic target policy, and specifically includes: if the feasible solution meeting the final deadline requirement does not exist during the first-generation initialization, the task flow execution time is taken as an optimization target, the adaptive value of the chromosome is evaluated by the task flow execution time in a reciprocal manner, the chromosome with smaller TET and higher adaptive value is selected in each iteration until the solution obtained by the chromosome appearing in the population meets the final deadline requirement, and the adaptive value is evaluated by the reciprocal of the task flow execution cost again;
in the embodiment, the selection operator module is used for selecting chromosome individuals entering the next generation through roulette according to the fitness value of the chromosome;
in this embodiment, the chromosome crossing and mutation module is configured to complete crossing and mutation of chromosomes, each chromosome is provided with a crossing probability and a mutation probability, a random number is generated for each resource scheduling sequence, and if the random number is smaller than the crossing probability, the resource scheduling sequence is selected as one of male parents of the crossing;
generating a random number for each task of each resource scheduling sequence, wherein if the random number is less than the variation probability, the resource number occupied by the task is varied;
in this embodiment, the optimal individual retention policy construction module is configured to construct an optimal individual retention policy, and specifically includes: in each generation, if the adaptive value of a chromosome in the current generation is higher than that of the current optimal chromosome, taking the chromosome in the current generation as a new optimal chromosome; otherwise, replacing the chromosome with the minimum fitness value in the current generation with the current optimal chromosome individual;
in this embodiment, the output module is configured to output the optimal solution, and after finding a feasible solution, when the feasible solution cannot be found after the specified algebra is performed, output the solution of the generated optimal chromosome.
Example 3
The present embodiment further provides a storage medium, where the storage medium may be a storage medium such as a ROM, a RAM, a magnetic disk, an optical disk, and the like, and the storage medium stores one or more programs, and when the programs are executed by a processor, the cloud computing resource scheduling optimization method based on the dynamic objective policy in embodiment 1 above is implemented.
Example 4
The embodiment also provides a computing device, where the computing device may be a desktop computer, a notebook computer, a smart phone, a PDA handheld terminal, a tablet computer, or other terminal devices with a display function, the computing device includes a processor and a memory, the memory stores one or more programs, and when the processor executes the programs stored in the memory, the cloud computing resource scheduling optimization method based on the dynamic objective policy in embodiment 1 is implemented.
The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.

Claims (7)

1. A cloud computing resource scheduling optimization method based on a dynamic objective strategy is characterized by comprising the following steps:
initializing a population: randomly generating a resource scheduling sequence for each chromosome in the population;
and (3) evaluating an adaptive value: evaluating an adaptive value by adopting the reciprocal of the task flow execution cost;
dynamic target policy: if the feasible solution meeting the final deadline requirement does not exist during the first-generation initialization, the task flow execution time is taken as an optimization target, the adaptive value of the chromosome is evaluated by the task flow execution time in a reciprocal manner, the chromosome with smaller TET and higher adaptive value is selected in each iteration until the solution obtained by the chromosome appearing in the population meets the final deadline requirement, and the adaptive value is evaluated by the reciprocal of the task flow execution cost again;
selecting an operator: selecting chromosome individuals entering the next generation by roulette according to the fitness value of the chromosomes;
crossover and mutation of chromosomes: each chromosome is provided with a cross probability and a variation probability, a random number is generated for each resource scheduling sequence, and if the random number is smaller than the cross probability, the resource scheduling sequence is selected as one of the male parents of the cross;
generating a random number for each task of each resource scheduling sequence, wherein if the random number is less than the variation probability, the resource number occupied by the task is varied;
optimal individual retention strategy: in each generation, if the adaptive value of a chromosome in the current generation is higher than that of the current optimal chromosome, taking the chromosome in the current generation as a new optimal chromosome; otherwise, replacing the chromosome with the minimum fitness value in the current generation with the current optimal chromosome individual;
and when a feasible solution cannot be found after the feasible solution is found out and the specified algebra is executed, outputting the solution of the generated optimal chromosome.
2. The dynamic objective policy-based cloud computing resource scheduling optimization method according to claim 1, wherein a resource scheduling sequence is randomly generated for each chromosome in the population, and the specific steps include:
the corresponding resources and tasks are represented by the numbers of the resources and the tasks, and the ith gene of each chromosome represents the task tiResource being executed, for chromosome x, xiJ denotes task tiIn resource rjIs executed.
3. The dynamic objective policy-based cloud computing resource scheduling optimization method according to claim 1, wherein the individual chromosome entering the next generation is selected by roulette, and the probability of the resource scheduling sequence being selected is:
Figure FDA0002723651910000021
fitnessi＝1/TECi
wherein TEC represents the task flow execution cost, fitnessiIndicates the fitness value, and i indicates the ith position of the chromosomal gene.
4. The dynamic objective policy-based cloud computing resource scheduling optimization method according to claim 1, wherein in the step of crossing and mutating chromosomes, each time 2 chromosomes are selected, a random number N e [1, N-1], N e Z, corresponding to the resource numbers of the first N tasks of two resource scheduling sequences are exchanged.
5. A cloud computing resource scheduling optimization system based on a dynamic objective strategy is characterized by comprising: the system comprises a population initialization module, an adaptive value evaluation module, a dynamic target strategy construction module, a selection operator module, a chromosome crossing and mutation module, an optimal individual retention strategy construction module and an output module;
the population initialization module is used for initializing a population and randomly generating a resource scheduling sequence for each chromosome in the population;
the adaptive value evaluation module is used for evaluating an adaptive value by adopting the reciprocal of the task flow execution cost;
the dynamic target strategy construction module is used for constructing a dynamic target strategy, and specifically comprises the following steps: if the feasible solution meeting the final deadline requirement does not exist during the first-generation initialization, the task flow execution time is taken as an optimization target, the adaptive value of the chromosome is evaluated by the task flow execution time in a reciprocal manner, the chromosome with smaller TET and higher adaptive value is selected in each iteration until the solution obtained by the chromosome appearing in the population meets the final deadline requirement, and the adaptive value is evaluated by the reciprocal of the task flow execution cost again;
the selection operator module is used for selecting chromosome individuals entering the next generation through roulette according to the adaptive value of the chromosomes;
the chromosome crossing and mutation module is used for completing the crossing and mutation of chromosomes, each chromosome is provided with a crossing probability and a mutation probability, a random number is generated for each resource scheduling sequence, and if the random number is smaller than the crossing probability, the resource scheduling sequence is selected as one of the male parents of the crossing;
generating a random number for each task of each resource scheduling sequence, wherein if the random number is less than the variation probability, the resource number occupied by the task is varied;
the optimal individual retention strategy construction module is used for constructing an optimal individual retention strategy, and specifically comprises the following steps: in each generation, if the adaptive value of a chromosome in the current generation is higher than that of the current optimal chromosome, taking the chromosome in the current generation as a new optimal chromosome; otherwise, replacing the chromosome with the minimum fitness value in the current generation with the current optimal chromosome individual;
the output module is used for outputting an optimal solution, and outputting the generated optimal chromosome solution when a feasible solution cannot be found after a specified algebra is found.
6. A storage medium storing a program, wherein the program, when executed by a processor, implements the method for cloud computing resource scheduling optimization based on dynamic objective policy according to any one of claims 1 to 4.
7. A computing device comprising a processor and a memory for storing processor-executable programs, wherein the processor, when executing a program stored in the memory, implements the dynamic objective policy-based cloud computing resource scheduling optimization method of any one of claims 1-4.
CN202011095662.3A 2020-10-14 2020-10-14 Cloud computing resource scheduling optimization method based on dynamic target strategy and application Pending CN112379996A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011095662.3A CN112379996A (en) 2020-10-14 2020-10-14 Cloud computing resource scheduling optimization method based on dynamic target strategy and application

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011095662.3A CN112379996A (en) 2020-10-14 2020-10-14 Cloud computing resource scheduling optimization method based on dynamic target strategy and application

Publications (1)

Publication Number Publication Date
CN112379996A true CN112379996A (en) 2021-02-19

Family

ID=74581444

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011095662.3A Pending CN112379996A (en) 2020-10-14 2020-10-14 Cloud computing resource scheduling optimization method based on dynamic target strategy and application

Country Status (1)

Country Link
CN (1) CN112379996A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114356564A (en) * 2021-12-29 2022-04-15 四川大学 System for integrating service resources
CN116755393A (en) * 2023-05-06 2023-09-15 成都飞机工业（集团）有限责任公司 Large-scale flexible job shop scheduling method, system, equipment and medium
CN117252395A (en) * 2023-11-10 2023-12-19 南京信息工程大学 Double-chromosome genetic algorithm-based multi-logistics vehicle scheduling method with service constraint

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110082821A1 (en) * 2009-10-07 2011-04-07 Mohammad Ali Abido Method of generating precedence-preserving crossover and mutation operations in genetic algorithms
CN102662756A (en) * 2012-02-27 2012-09-12 北京航空航天大学 Resource-adjustable satellite parallel-testing resource configuration method
CN103902375A (en) * 2014-04-11 2014-07-02 北京工业大学 Cloud task scheduling method based on improved genetic algorithm
CN108880663A (en) * 2018-07-20 2018-11-23 大连大学 Incorporate network resource allocation method based on improved adaptive GA-IAGA
CN109800071A (en) * 2019-01-03 2019-05-24 华南理工大学 A kind of cloud computing method for scheduling task based on improved adaptive GA-IAGA
WO2019217573A1 (en) * 2018-05-08 2019-11-14 Vmware, Inc. Task assignment in virtual gpu enabled systems

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110082821A1 (en) * 2009-10-07 2011-04-07 Mohammad Ali Abido Method of generating precedence-preserving crossover and mutation operations in genetic algorithms
CN102662756A (en) * 2012-02-27 2012-09-12 北京航空航天大学 Resource-adjustable satellite parallel-testing resource configuration method
CN103902375A (en) * 2014-04-11 2014-07-02 北京工业大学 Cloud task scheduling method based on improved genetic algorithm
WO2019217573A1 (en) * 2018-05-08 2019-11-14 Vmware, Inc. Task assignment in virtual gpu enabled systems
CN108880663A (en) * 2018-07-20 2018-11-23 大连大学 Incorporate network resource allocation method based on improved adaptive GA-IAGA
CN109800071A (en) * 2019-01-03 2019-05-24 华南理工大学 A kind of cloud computing method for scheduling task based on improved adaptive GA-IAGA

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
ZHI-HUI ZHAN 等: "A Distributed Multiple Populations Framework for Evolutionary Algorithm in Solving Dynamic Optimization Problems", 《IEEEACCESS》 *
ZHI-HUI ZHAN 等: "A Distributed Multiple Populations Framework for Evolutionary Algorithm in Solving Dynamic Optimization Problems", 《IEEEACCESS》, 19 March 2019 (2019-03-19), pages 1 - 19 *
ZONG-GAN CHEN ET AL.: "Deadline constrained cloud computing resources scheduling for cost optimization based on dynamic objective genetic algorithm", 2015 IEEE CONGRESS ON EVOLUTIONARY COMPUTATION（CEC）, pages 708 - 714 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114356564A (en) * 2021-12-29 2022-04-15 四川大学 System for integrating service resources
CN116755393A (en) * 2023-05-06 2023-09-15 成都飞机工业（集团）有限责任公司 Large-scale flexible job shop scheduling method, system, equipment and medium
CN117252395A (en) * 2023-11-10 2023-12-19 南京信息工程大学 Double-chromosome genetic algorithm-based multi-logistics vehicle scheduling method with service constraint
CN117252395B (en) * 2023-11-10 2024-02-06 南京信息工程大学 Double-chromosome genetic algorithm-based multi-logistics vehicle scheduling method with service constraint

Similar Documents

Publication Publication Date Title
CN112379996A (en) Cloud computing resource scheduling optimization method based on dynamic target strategy and application
CN108829501B (en) Batch processing scientific workflow task scheduling algorithm based on improved genetic algorithm
CN108664330B (en) cloud resource allocation method based on variable neighborhood search strategy
Cui et al. Workflow tasks scheduling optimization based on genetic algorithm in clouds
CN107133091A (en) The cloud workflow task dispatching method being classified based on top-down task
CN107656799B (en) Workflow scheduling method considering communication and computing cost under multi-cloud environment
CN110008023B (en) Cloud computing system budget constraint random task scheduling method based on genetic algorithm
CN107862404B (en) Manufacturing service supply chain optimization method based on service correlation
Shishira et al. Survey on meta heuristic optimization techniques in cloud computing
Pourebrahimi et al. Market-based resource allocation in grids
CN110162390B (en) Task allocation method and system for fog computing system
CN113821318A (en) Internet of things cross-domain subtask combined collaborative computing method and system
CN111813500B (en) Multi-target cloud workflow scheduling method and device
CN112685138A (en) Multi-workflow scheduling method based on multi-population hybrid intelligent optimization in cloud environment
Deniziak et al. Cost optimization of real-time cloud applications using developmental genetic programming
CN110851247A (en) Cost optimization scheduling method for constrained cloud workflow
CN114297934A (en) Model parameter parallel simulation optimization method and device based on proxy model
CN110928648B (en) Heuristic and intelligent computing-fused cloud workflow segmentation online scheduling optimization method
CN112380016A (en) Cloud computing resource load balancing scheduling method based on improved genetic algorithm and application
CN110928670B (en) Workflow optimization method based on partial sequence adaptive genetic algorithm in cloud computing environment
CN114980216A (en) Dependent task unloading system and method based on mobile edge calculation
CN112306642B (en) Workflow scheduling method based on stable matching game theory
CN110321208B (en) Evolutionary computing method for solving cloud task scheduling
CN113220437A (en) Workflow multi-target scheduling method and device
Alkhanak et al. A hyper-heuristic approach using a prioritized selection strategy for workflow scheduling in cloud computing

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination