CN108320059A - A kind of workflow schedule evolution optimization method and terminal device - Google Patents
A kind of workflow schedule evolution optimization method and terminal device Download PDFInfo
- Publication number
- CN108320059A CN108320059A CN201810154127.7A CN201810154127A CN108320059A CN 108320059 A CN108320059 A CN 108320059A CN 201810154127 A CN201810154127 A CN 201810154127A CN 108320059 A CN108320059 A CN 108320059A
- Authority
- CN
- China
- Prior art keywords
- chromosome
- fitness
- scheduling
- population
- strategy
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/10—Office automation; Time management
- G06Q10/103—Workflow collaboration or project management
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- General Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Data Mining & Analysis (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Genetics & Genomics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Physiology (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention is suitable for cloud computing and field of artificial intelligence, provides a kind of workflow schedule evolution optimization method and terminal device.The workflow schedule evolution optimization method includes：Multiple chromosome is designed to coding structure；The balanced mixing initialization population of construction；Adaptive polo placement correlated variables；Execute evolutional operation again；It is preferred that scheduling strategy.Above-mentioned workflow schedule evolution optimization method, can with higher deadline constraint satisfaction rate and it is relatively low execute cost, the scheduling strategy for meeting user demand is provided, is provided with reference to realization for the lower workflow schedule evolution optimization system of structure deadline constraint.
Description
Technical field
The invention belongs to cloud computing and field of artificial intelligence more particularly to a kind of workflow schedule evolution optimization methods
And terminal device.
Background technology
Task in cloud computing exists in the form of workflow more and more.This workflow be typically it is data-intensive and
Compute-intensive applications.Since workflow has a large amount of data and calculates demand, so cloud computing is needed to provide high-performance meter
Calculate resource.Cloud computing is dynamically supplied to user in the form of virtual machine according to the demand of user using computing resource as service.
Cloud computing provide resource service pattern include mainly：Infrastructure services (IaaS), platform services (PaaS) and software
Service (SaaS).Service is provided to the user in a manner of virtualization pool in infrastructure cloud, is work by dispatching technique
Stream distribution computing resource is as cloud computing, the research emphasis of artificial intelligence field.
Under cloud environment workflow schedule refer under specific cloud environment, according to the calculating demand of user, be in workflow not
Same calculating task matches different computing resources, and searches out the process of Optimum Matching scheme.To task when workflow schedule
Resource allocation and optimizing are carried out, by extracting workflow task feature and mutual data dependence relation, obtains workflow issues
Formulation description, then it is modeled, computing resource is distributed for it, determines that allocation strategy meets user demand, and optimizing.Work
Make stream scheduling to attempt to establish the correspondence between calculating task, computing resource and user demand, measures its adaptedness.
Scheduling is used for the mapping between calculating task and computing resource, such as computational by the calculating task and difference of different calculation scales
The time is executed between the computing resource of energy and executes the variation spent, finds out the track that objective optimization item changes.In cloud computing work
Make in stream scheduling system, workflow content is usually described with node diagnostic.In fact, the workflow based on cloud computing can be with
It is divided into three steps：Workflow modeling, task resource mapping and optimal policy finding.
Workflow schedule is an important application of artificial intelligence, and main purpose is for distribution of computation tasks institute in workflow
It needs computing resource and is adjusted, the optimization problem of scheduling of resource alternatively referred to as under constraints.Scheduling of resource is mainly to meter
Computing resource in resource pool is calculated to be allocated.In general, the processing of scheduling of resource optimization problem is to use penalty under constraints
It punishes inappropriate scheduling strategy, restricted problem is converted into unconstrained problem.In addition to conventional scheduling algorithms continue to optimize and
Except improvement, some evolution algorithms are also employed to generate the scheduling strategy of near-optimization.Rodriguez and Buyya is directed to cloud
Scientific workflow scheduling proposes a kind of particle swarm optimization algorithm inspiring optimisation technique based on member under environment, which exists
Meeting makes overall execution least cost under deadline constraints, some for having fully considered IaaS in cloud computing are substantially special
Property, such as the elastification and isomerization of computing resource.But using being unsatisfactory for the particle of constraints not as good as legal particle in algorithm
Penalty may result in algorithm Premature Convergence and be absorbed in local optimum.When Li Liu et al. people propose one kind based on cut-off
Between the collaboration genetic algorithm CGA that constrains2, ensure that in the case where meeting deadline constraints, make overall execution least cost, energy
It is enough to accelerate to restrain and be avoided that precocity.But coevolution operation is introduced in the algorithm and will increase calculation amount, simultaneously because it is essential
Optimal value can not can be obtained by constraint demand in some cases by being a kind of Greedy strategy.
Invention content
In view of this, an embodiment of the present invention provides a kind of workflow schedule evolution optimization method and terminal device, with solution
Certainly deadline constraint satisfaction rate is relatively low in the prior art and executes the higher problem of cost.
The embodiment of the present invention in a first aspect, provide a kind of workflow schedule evolution optimization method, including：
Step S1 chooses chromosome in Population in Genetic Algorithms and is used as scheduling according to user's input parameter and Work stream data
Policy information carrier designs information of the multiple chromosome to W in conjunction with population scale M, computing resource scale N and Work stream data
Coding structure；User's input parameter includes deadline constrained parameters；
Step S2 is recycled single according to any multiple chromosome pair that described information coding structure constructs according to building method
A a variety of chromosomes form the balanced mixing initialization population that scale is M to construction process；
Step S3 builds evolutionary process, and according to user's input parameter according to the balanced mixing initialization population
Evolutionary direction is adjusted, according to user's input parameter and original fitness amendment mean value construction improvement fitness, and according to
The fitness calculates the adaptive crossover mutation Pc and self-adaptive mutation Pm of every chromosome of population in evolutionary process, will
Every chromosome carries out genetic manipulation and gained scheduling strategy is stored in scheduling optimizing strategy pond A；
Step S4, the input of evolutional operation again under ω chromosome in the A of preference policy pond is constrained as deadline,
And the chromosome that M- ω items generate at random is added, Advanced group species are obtained again as the input of step S3 and according to established rule iteration
Execution is evolved again, and gained scheduling strategy is stored in scheduling optimizing strategy pond B；
Step S5, overall merit dispatch optimizing strategy pond A and dispatch every scheduling strategy of optimizing strategy pond B, and matching is cut
Only time constraint condition, and compare deadline constraint satisfaction rate and execute cost, evaluation algorithms execution efficiency finally judges excellent
Selection scheduling strategy.
Optionally, the multiple chromosome in step S1 includes 2 chromosomes altogether to the information coding structure of W, i.e.,
Multiple chromosome is to W={ chromesome1, chromesome2 }
Chromesome1=sequence of { Pc, Pm, gen0,gen1,…,genk,…,genM-1}
Chromesome2={ tof0,tof1,…,tofi,…,tofN-1}
Assuming that on i-th virtual machine, the scale for executing sequence is g, then tofiIt is represented by：
Wherein, Pc is the adaptive crossover mutation of chromosome；Pm is the self-adaptive mutation of chromosome；genkFor dyeing
Gene on the k location of body, 0≤k<M；vmiFor No. i-th virtual machine, 0≤i in computing resource<N； tofiMoney is calculated to be assigned to
Source vmiCalculating task execute sequence；Gene gen's is encoded to t (vm), expression calculating task in chromesome1 chromosomes
T distributes computing resource vm；Gene vm indicates that computing resource, subsequent gene set { t } indicate to divide in chromesome2 chromosomes
Fit over the calculating task of computing resource vm.
Optionally, any multiple chromosome pair constructed according to described information coding structure recycles single according to building method
A a variety of chromosomes form the balanced mixing initialization population that scale is M to construction process, and detailed process is as follows：
The chromosome obtained according to step S1 is to coding structure, it is assumed that use k kind method construct initialization populations, then according to
Method 1 constructs x1Chromosome pair constructs x according to method iiChromosome pair constructs x according to method kkChromosome pair, obtains
The balanced mixing initial population for being M to scale；Wherein, k >=2,
Optionally, it corrects mean value construction according to user's input parameter and original fitness and improves fitness, and root
The adaptive crossover mutation Pc and self-adaptive mutation Pm of every chromosome of population in evolutionary process are calculated according to the fitness,
Detailed process is as follows：
For the balanced mixing initialization population, the original fitness vector for calculating chromosome is constrained according to deadline
Fini, mean value is corrected by original fitness, construction improves fitness vector Fimp；
Using greatest improvement fitness value subtract it is current improve fitness value, and divided by greatest improvement fitness value with improve
Fitness corrects the difference of mean value, obtains probability intermediate variable Vmi；
When being less than its amendment mean value to improving fitness value, it is 1 that probability, which corrects variable-value,；
The relationship of mean value is corrected with it to improving fitness value, is multiplied respectively with different parameters, is labeled as Pc；
M target chromosome sample is taken, above procedure is repeated and obtains M Pc value；
Based on same process, Pm can be obtained.
Optionally, judge that the process of preferred scheduling strategy is specific as follows in step S5：
Step S2, step S3 are executed to all chromosomes in population and step S4 obtains scheduling optimizing strategy pond A and tune
Optimizing strategy pond B is spent, using deadline constraint satisfaction rate as evaluation criterion, compares the obtained scheduling optimizing of step S4 and step S5
Tactful pond A and scheduling optimizing strategy pond B；
According to user's input information, secondary constraint is determined；User's input information further includes secondary constraint information；
If a certain scheduling optimizing strategy fitness is optimal and does not violate secondary deadline constraint, or violates deadline about
Shu Chengdu is minimum with respect to other scheduling optimizing strategies, then judges the scheduling optimizing strategy as target dispatch optimizing strategy, and give
Go out scheduling result.
The second aspect of the embodiment of the present invention provides a kind of terminal device, including memory, processor, the storage
The computer program that can be run on the processor is stored in device, the processor is realized when executing the computer program
Following steps：
Step S1 chooses chromosome in Population in Genetic Algorithms and is used as scheduling according to user's input parameter and Work stream data
Policy information carrier designs information of the multiple chromosome to W in conjunction with population scale M, computing resource scale N and Work stream data
Coding structure；User's input parameter includes deadline constrained parameters；
Step S2 is recycled single according to any multiple chromosome pair that described information coding structure constructs according to building method
A a variety of chromosomes form the balanced mixing initialization population that scale is M to construction process；
Step S3 builds evolutionary process, and according to user's input parameter according to the balanced mixing initialization population
Evolutionary direction is adjusted, according to user's input parameter and original fitness amendment mean value construction improvement fitness, and according to
The fitness calculates the adaptive crossover mutation Pc and self-adaptive mutation Pm of every chromosome of population in evolutionary process, will
Every chromosome carries out genetic manipulation and gained scheduling strategy is stored in scheduling optimizing strategy pond A；
Step S4, the input of evolutional operation again under ω chromosome in the A of preference policy pond is constrained as deadline,
And the chromosome that M- ω items generate at random is added, Advanced group species are obtained again as the input of step S3 and according to established rule iteration
Execution is evolved again, and gained scheduling strategy is stored in scheduling optimizing strategy pond B；
Step S5, overall merit dispatch optimizing strategy pond A and dispatch every scheduling strategy of optimizing strategy pond B, and matching is cut
Only time constraint condition, and compare deadline constraint satisfaction rate and execute cost, evaluation algorithms execution efficiency finally judges excellent
Selection scheduling strategy.
Optionally, the multiple chromosome in step S1 includes 2 chromosomes altogether to the information coding structure of W, i.e.,
Multiple chromosome is to W={ chromesome1, chromesome2 }
Chromesome1=sequence of { Pc, Pm, gen0,gen1,…,genk,…,genM-1}
Chromesome2={ tof0,tof1,…,tofi,…,tofN-1}
Assuming that on i-th virtual machine, the scale for executing sequence is g, then tofiIt is represented by：
Wherein, Pc is the adaptive crossover mutation of chromosome；Pm is the self-adaptive mutation of chromosome；genkFor dyeing
Gene on the k location of body, 0≤k<M；vmiFor No. i-th virtual machine, 0≤i in computing resource<N； tofiMoney is calculated to be assigned to
Source vmiCalculating task execute sequence；Gene gen's is encoded to t (vm), expression calculating task in chromesome1 chromosomes
T distributes computing resource vm；Gene vm indicates that computing resource, subsequent gene set { t } indicate to divide in chromesome2 chromosomes
Fit over the calculating task of computing resource vm.
Optionally, any multiple chromosome pair constructed according to described information coding structure recycles single according to building method
A a variety of chromosomes form the balanced mixing initialization population that scale is M to construction process, and detailed process is as follows：
The chromosome obtained according to step S1 is to coding structure, it is assumed that use k kind method construct initialization populations, then according to
Method 1 constructs x1Chromosome pair constructs x according to method iiChromosome pair constructs x according to method kkChromosome pair, obtains
The balanced mixing initial population for being M to scale；Wherein, k >=2,
Optionally, it corrects mean value construction according to user's input parameter and original fitness and improves fitness, and root
The adaptive crossover mutation Pc and self-adaptive mutation Pm of every chromosome of population in evolutionary process are calculated according to the fitness,
Detailed process is as follows：
For the balanced mixing initialization population, the original fitness vector for calculating chromosome is constrained according to deadline
Fini, mean value is corrected by original fitness, construction improves fitness vector Fimp；
Using greatest improvement fitness value subtract it is current improve fitness value, and divided by greatest improvement fitness value with improve
Fitness corrects the difference of mean value, obtains probability intermediate variable Vmi；
When being less than its amendment mean value to improving fitness value, it is 1 that probability, which corrects variable-value,；
The relationship of mean value is corrected with it to improving fitness value, is multiplied respectively with different parameters, is labeled as Pc；
M target chromosome sample is taken, above procedure is repeated and obtains M Pc value；
Based on same process, Pm can be obtained.
Optionally, judge that the process of preferred scheduling strategy is specific as follows in step S5：
Step S2, step S3 are executed to all chromosomes in population and step S4 obtains scheduling optimizing strategy pond A and tune
Optimizing strategy pond B is spent, using deadline constraint satisfaction rate as evaluation criterion, compares the obtained scheduling optimizing of step S4 and step S5
Tactful pond A and scheduling optimizing strategy pond B；
According to user's input information, secondary constraint is determined；User's input information further includes secondary constraint information；
If a certain scheduling optimizing strategy fitness is optimal and does not violate secondary deadline constraint, or violates deadline about
Shu Chengdu is minimum with respect to other scheduling optimizing strategies, then judges the scheduling optimizing strategy as target dispatch optimizing strategy, and give
Go out scheduling result.
The third aspect of the embodiment of the present invention, provides a kind of computer readable storage medium, described computer-readable to deposit
Storage media is stored with computer program, and such as first aspect of the embodiment of the present invention is realized when the computer program is executed by processor
The step of described either method.
The embodiment of the present invention the characteristics of for workflow task set, proposes that one kind is evolved (Re- again based on genetic algorithm
Evolution) strategy solves the problems, such as that local optimum and convergence rate are slow, and proposes that one kind of multiple chromosomes tie coding
Structure can fully describe workflow schedule strategy according to workflow task nodal information, and then pass through adaptive polo placement deadline
Constraint is lower to correct the correlated variables of mean value to ensure the survival rate of effective gene, according to set from two scheduling optimizing resource pools
Regular preferably final scheduling strategy.It is compared with the existing dispatching method based on evolution algorithm, the method reduce calculation amounts, can protect
The Service Efficiency that specific application context restrictions need is demonstrate,proved, and ensures the controllable execution efficiency of algorithm, can avoid local optimum and convergence
The problems such as speed is slow.
Description of the drawings
It to describe the technical solutions in the embodiments of the present invention more clearly, below will be to embodiment or description of the prior art
Needed in attached drawing be briefly described, it should be apparent that, the accompanying drawings in the following description be only the present invention some
Embodiment for those of ordinary skill in the art without having to pay creative labor, can also be according to these
Attached drawing obtains other attached drawings.
Fig. 1 is workflow schedule evolution optimization method block diagram provided in an embodiment of the present invention；
Fig. 2 is workflow schedule evolution optimization method flow chart provided in an embodiment of the present invention；
Fig. 3 is scientific workflow structural schematic diagram provided in an embodiment of the present invention；
Fig. 4 is the running environment schematic diagram of workflow schedule evolution optimizing program provided in an embodiment of the present invention；
Fig. 5 is the Program modual graph of workflow schedule evolution optimizing program provided in an embodiment of the present invention.
Specific implementation mode
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation describes, it is clear that described embodiments are some of the embodiments of the present invention, instead of all the embodiments.Based on this hair
Embodiment in bright, every other implementation obtained by those of ordinary skill in the art without making creative efforts
Example, shall fall within the protection scope of the present invention.
The present invention be directed to workflow task set, structure it is a kind of based on genetic algorithm (Genetic Algorithm,
GA workflow schedule evolution optimization method), design description multiple chromosome establish efficient chromosome pair to coding structure,
The balanced mixing initialization population of construction, correlated variables of the adaptive polo placement based on contemporary population's fitness design evolutional operation again,
Accelerate to provide with reference to realization for workflow schedule optimal policy finding and convergence.
The present invention is the deadline constraint provided according to user, adaptive polo placement correlated variables, structure evolutionary process with
Evolutionary process again, Comprehensive Assessment scheduling strategy is good and bad, and the computational methods of optimizing final result.For in workflow task set
The information characteristics of task node, it is proposed that multiple chromosome is to coding structure, balanced mixing initialization population, calculating deadline
The correlated variables of the lower adaptive correction mean value of constraint, again evolution strategy, scheduling optimizing Policies Resource pond and optimizing scheduling strategy etc.
Concept devises the workflow schedule evolution optimization method under a kind of new deadline constraint.
In order to illustrate technical solutions according to the invention, illustrated below by specific embodiment.
In order to accurately illustrate the embodiment of the present invention, it is explained as follows term and meaning first.
Task：Refer to the basic unit scheduled in scheduling evolution optimization method.Gene table in chromosome is used in the present invention
Show task.Task has mission number, scale, receives the information such as data scale, transmission data scale and dependence.
Resource：Resource is used for calculating task, refers in particular to virtual machine in the present invention.Usually, virtual machine is compiled with virtual machine
Number, the information such as performance, state and price.
Workflow：A kind of form of set of tasks.Usually, workflow is the collection comprising multiple tasks and its dependence
It closes, wherein scientific workflow is one kind of more important complexity in workflow.
Chromosome pair：Chromosome to refering in particular to designed multiple chromosome pair in the present invention, for describing the master of work
Feature, the different chromosomes of chromosome centering is wanted to play different role in the different genes operational phase, it is also known as individual.
Adaptive polo placement correlated variables：Adaptive dependent variable in evolution optimization method is dispatched, fitness, crossover probability are such as improved
With mutation probability, select probability and penalty etc..Computing resource isomerism is can avoid by adaptive polo placement correlated variables
With precocious caused by the characteristics such as the dynamic and convergent deficiency in part.
Genetic manipulation：Genetic manipulation refers mainly to the operations such as intersection, variation and selection in evolutionary process of the present invention.
Dispatch optimizing Policies Resource pond：Assigned storage strategy provides support to evolve again with preferred scheduling strategy.
Constraint：Refer to the demand that user proposes, as main constraints.The present invention is using deadline as major constraints
Condition is spent with executing as secondary constraint.
Constraint satisfaction rate：Bound term and constraints are compared, is executed repeatedly several times, statistics meets the ratio of constraint, should
Ratio is known as constraint satisfaction rate, and constraint satisfaction rate is higher, and corresponding scheduling strategy is more outstanding.
Fitness：The quality for judging scheduling strategy is the key that determine Evolutionary direction.
It evolves again：On the basis of when evolution acquired results, Advanced group species again are rebuild according to established rule, again into
Change, the corresponding scheduling optimizing Policies Resource pond of the result that will evolve again deposit, can avoid being absorbed in common evolution result local optimum with
And precocious problem.
Embodiment one
Fig. 1 and Fig. 2 respectively illustrates the basic structure of the workflow schedule evolution optimization method of the offer of the embodiment of the present invention one
At with implementation process, details are as follows：
Step S1 chooses chromosome in Population in Genetic Algorithms and is used as scheduling according to user's input parameter and Work stream data
Policy information carrier designs information of the multiple chromosome to W in conjunction with population scale M, computing resource scale N and Work stream data
Coding structure；User's input parameter includes deadline constrained parameters.
In this step, multiple chromosome includes in different chromosomes to containing multiple chromosomes pair in coding structure
Different information.Work stream data is eliminated the reliance on during the scheduling evolution optimization method that the present invention announces, and uses multiple dye
Colour solid indicates task and relevant information in workflow to description.The method that the present invention announces is chosen being extracted from workflow for task and is compiled
Number (0- (M-1)), task quantity (M)；Number (0- (N-1)), the quantity (N) of virtual machine in resource pool；Crossover probability and variation
The information such as probability, design multiple chromosome is to W coding structures.
Wherein, the multiple chromosome in this step includes 2 chromosomes altogether to the information coding structure of W, i.e.,
Multiple chromosome is to W={ chromesome1, chromesome2 }
Chromesome1=sequence of { Pc, Pm, gen0,gen1,…,genk,…,genM-1}
Chromesome2={ tof0,tof1,…,tofi,…,tofN-1}
Assuming that on i-th virtual machine, the scale for executing sequence is g, then tofiIt is represented by：
Wherein, Pc is the adaptive crossover mutation of chromosome；Pm is the self-adaptive mutation of chromosome；genkFor dyeing
Gene on the k location of body, 0≤k<M；vmiFor No. i-th virtual machine, 0≤i in computing resource<N； tofiMoney is calculated to be assigned to
Source vmiCalculating task execute sequence；Gene gen's is encoded to t (vm), expression calculating task in chromesome1 chromosomes
T distributes computing resource vm；Gene vm indicates that computing resource, subsequent gene set { t } indicate to divide in chromesome2 chromosomes
Fit over the calculating task of computing resource vm.
Step S2 is recycled single according to any multiple chromosome pair that described information coding structure constructs according to building method
A a variety of chromosomes form the balanced mixing initialization population that scale is M to construction process.
As a kind of embodiment, the detailed process of this step is as follows：
The chromosome obtained according to step S1 is to coding structure, it is assumed that use k kind method construct initialization populations, then according to
Method 1 constructs x1Chromosome pair constructs x according to method iiChromosome pair constructs x according to method kkChromosome pair, obtains
The balanced mixing initial population for being M to scale；Wherein, k >=2,If according to the proposed coding structure structure of step (1)
Any multiple chromosome is made to a；According to a variety of building method dyeing cycle bodies to construction process, equilibrium mixing initialization population is
B.If population B has M chromosome pair, if shared k kind methods；The chromosome that method 1 constructs is combined into B1 to collection；Method 2 constructs
Chromosome to collection be combined into B2；…；The chromosome of method k constructions is combined into Bk to collection；Then initialization population B is by k Ge Zizhong groups
At.X is constructed according to method 11Chromosome pair, method 2 construct x2Chromosome is right ..., and method i constructs xiChromosome
Pair ..., method k construct xkChromosome pair, the balanced mixing initial population that the scale that obtains is M.
The chromosome that equipment, method 1 is constituted is to set
The chromosome that method 2 is constituted is to set
And so on, the chromosome that method k is constituted is combined into collection
Therefore, the composition of balanced mixing initialization population B is：B=B1, B2 ..., Bk }.
Step S3 builds evolutionary process, and according to user's input parameter according to the balanced mixing initialization population
Evolutionary direction is adjusted, according to user's input parameter and original fitness amendment mean value construction improvement fitness, and according to
The fitness calculates the adaptive crossover mutation Pc and self-adaptive mutation Pm of every chromosome of population in evolutionary process, will
Every chromosome carries out genetic manipulation and gained scheduling strategy is stored in scheduling optimizing strategy pond A.
It is optionally, described to correct mean value construction improvement fitness according to user's input parameter and original fitness,
And it is general according to the adaptive crossover mutation Pc of every chromosome of population and TSP question in fitness calculating evolutionary process
Rate Pm, detailed process are as follows：
For the balanced mixing initialization population, the original fitness vector for calculating chromosome is constrained according to deadline
Fini, mean value is corrected by original fitness, construction improves fitness vector Fimp；
Using greatest improvement fitness value subtract it is current improve fitness value, and divided by greatest improvement fitness value with improve
Fitness corrects the difference of mean value, obtains probability intermediate variable Vmi；
When being less than its amendment mean value to improving fitness value, it is 1 that probability, which corrects variable-value,；
The relationship of mean value is corrected with it to improving fitness value, is multiplied respectively with different parameters, is labeled as Pc；
M target chromosome sample is taken, above procedure is repeated and obtains M Pc value；
Based on same process, Pm can be obtained.
Specifically,
Wherein, Pc(i) crossover probability of i-th chromosome pair in contemporary population, P are indicatedm(i) it indicates i-th in contemporary population
The mutation probability of chromosome pair.In formula (1) and (2), when i-th chromosome is less than or equal to current population to fitness f (i)
The amendment mean value f of fitnesscavgWhen, it is believed that the chromosome makes Maturing the and Matured stages in evolutionary process
With cross parameter k1With Mutation parameter k2；When i-th chromosome is to fitness f'(i) it is repaiied more than or equal to current population's fitness
Positive mean value fcavgWhen, it is believed that the chromosome uses intersection to Initial the and Undermatured stages in evolutionary process
Parameter k3With Mutation parameter k4.In order to avoid being mutated the very poor chromosome generated to the influence to population entirety fitness average, originally
Embodiment corrects mean value f using λcavg, calculate as shown in formula (3), wherein λ corrects the percentage of shared population required for indicating
Than being only used for correcting very poor chromosome pair.
In formula (3), SpIndicate the size of current population, fc(i) indicate revised population chromosome to fitness.
Step S4, the input of evolutional operation again under ω chromosome in the A of preference policy pond is constrained as deadline,
And the chromosome that M- ω items generate at random is added, Advanced group species are obtained again as the input of step S3 and according to established rule iteration
Execution is evolved again, and gained scheduling strategy is stored in scheduling optimizing strategy pond B.
In this step, ω chromosome in optimizing Policies Resource pond A will be dispatched as the behaviour that evolves again according to established rule
The input of work, and the chromosome that (M- ω) item generates at random is added, input of the Advanced group species as step S3 again is obtained, according to both
Then evolutionary generation, gained chromosome are stored in scheduling optimizing Policies Resource pond B, scheduling strategy are provided for step S5 iteration set pattern again
Standby resources.
If it is S to dispatch the storage population in optimizing Policies Resource pond AA：SA={ a0,a1,...,aM-1}；
It is S according to the part population that established rule is chosen1：S1={ ai,ai+1,...ai+ω}；
Newly-generated part population is S2：S2={ a0,a1,...,aM-1-ω}；
Therefore, gained Advanced group species are Snew：
Snew={ S1,S2}
Step S5, overall merit dispatch optimizing strategy pond A and dispatch every scheduling strategy of optimizing strategy pond B, and matching is cut
Only time constraint condition, and compare deadline constraint satisfaction rate and execute cost, evaluation algorithms execution efficiency finally judges excellent
Selection scheduling strategy.
Optionally, judge that the process of preferred scheduling strategy is specific as follows in this step：
Step S2, step S3 are executed to all chromosomes in population and step S4 obtains scheduling optimizing strategy pond A and tune
Optimizing strategy pond B is spent, using deadline constraint satisfaction rate as evaluation criterion, compares the obtained scheduling optimizing of step S4 and step S5
Tactful pond A and scheduling optimizing strategy pond B；
According to user's input information, secondary constraint is determined；User's input information further includes secondary constraint information；
If a certain scheduling optimizing strategy fitness is optimal and does not violate secondary deadline constraint, or violates deadline about
Shu Chengdu is minimum with respect to other scheduling optimizing strategies, then judges the scheduling optimizing strategy as target dispatch optimizing strategy, and give
Go out scheduling result.
Specifically, setting number of evolving again as z；Formed population of evolving again is Sr；Dispatch the storage in optimizing strategy pond B
Cluster is combined into SB：SB={ Sr1,Sr2,...,Srz}；If the highest chromosome number of constraint satisfaction rate is h, then set expression
For MB：MB={ m0,m1,...,mh-1}；If it is e to execute and spend minimum chromosome number, then set expression is CB：CB=
{c0,c1,...,ce-1}。
Compare population set SBWith population SAIn each chromosome, first compare constraint satisfaction rate, find and wherein constrain
The highest all chromosomes of Service Efficiency, by these chromosomes deposit constraint satisfaction rate chromosome congression MB；If MBScale it is big
In 1, then set M is traversedBIn all chromosomes, then comparison executes costs, finds the minimum all chromosomes of execution cost, deposit
Enter to execute and spends chromosome congression CB；If CBScale be more than 1, then traverse set CBIn all chromosomes, then control methods effect
Rate finds the highest all chromosomes of efficiency, if chromosome item number is more than 1, randomly selects item chromosome as a result
Output.
Illustrate the process of the present embodiment with specific example below.
Experiment uses scientific workflow montage-25, wherein 25 task nodes, structural schematic diagram is as shown in Fig. 3；
And simulated using 5 different types of virtual machines, virtual machine parameter setting is with reference to Amazon EC2 configurations.
(1) design multiple chromosome is to coding structure
If multiple chromosome is set W to description workflow task.As space is limited, wherein the 0th chromosome pair is only enumerated
Data,
W0={ chromesome1, chromesome2 }
Chromesome1=sequence of { 0.45,0.011, T4(0),...T24(4)}
Chromesome2=sequence of { vm0,T4,T1,T12,T16,T20,T21,T23}
...
sequence of{vm4,T7,T10,T11,T17,T24}
Wherein, 2 genes 0.45,0.011 are respectively crossover probability and mutation probability before chromesome1；25 bases afterwards
Because of the mapping pair of calculating task and computing resource, totally 27 genes.
Equally, chromesome2 shares 5 sub- genomes at the 1st gene representation is distributed wherein in daughter chromosome
Computing resource, subsequent gene indicates the calculating task that executes on No. 1 gene.
(2) the balanced mixing initialization population of construction
The workflow task characteristic set W that multiple chromosome pair is obtained according to step (1) shares 50 chromosomes pair.Example
Such as, 2 kinds of dispatching algorithms and random algorithm are selected, each scheduling 15 chromosome pair of each generations, 20 chromosomes of residue to by with
Machine algorithm generates.
(3) adaptive polo placement correlated variables
By the crossover probability and mutation probability of all chromosomes pair in formula (1) and (2) adaptive polo placement W, probability is obtained
Matrix P,
And corresponding crossover probability and mutation probability are written to the corresponding gene position of homologue pair.
(4) it executes and evolves again
25 chromosomes in optimizing Policies Resource pond A will be dispatched to as the defeated of evolutional operation again according to established rule
Enter, and 25 chromosomes pair generated at random are added, input of the Advanced group species as step (3) again is obtained, according to established rule
Evolutionary generation, gained chromosome are standby for step (5) offer scheduling strategy to being stored in scheduling optimizing Policies Resource pond B again for iteration
Use resource.
If it is S to dispatch the storage population in optimizing Policies Resource pond AA：
SA={ a0,a1,...,a49}；
It is S according to the part population that established rule is chosen1：
S1={ ai,ai+1,...,ai+24}；
Newly-generated part population is S2：S2={ a0,a1,...,a24}；
Therefore, gained Advanced group species are Snew：
Snew={ S1,S2}
As space is limited, the data of wherein the 0th chromosome pair are only enumerated,
W0={ chromesome1, chromesome2 }
Chromesome1=sequence of { 0.25,0.002, T0(2),...,T24(3)}
Chromesome2=sequence of { vm0,T4,T1,T3,...,T14,T16,T22}
...
sequence of{vm4,T10,T20,T17,T19,T21,T23}
(5) preferred scheduling strategy
Overall merit dispatches optimizing strategy pond A and dispatches every chromosome in optimizing strategy pond B, comparison fitness value and
Cost is executed, matching constraint condition finally judges preferred scheduling strategy Final.
Final={ chromesome1, chromesome2 }
Chromesome1=sequence of { 0.07,0.001, T0(2),...,T24(2)}
Chromesome2=sequence of { vm0,T4,T1,T3,...,T14,T17,T21}
...
sequence of{vm4,T10,T20,T16,T19,T22}
The present embodiment and the Comparative result of other two methods are as shown in table 1, are such as constrained in application requirement deadline
Constraint satisfaction rate proposed by the present invention is 100% when 328.7 unit interval, executes that spend be that 13.2 units are spent；PSO methods
Constraint satisfaction rate is 10%, executes and spends as the cost of 41.6 units；CGA2The constraint satisfaction rate of method is 60%, executes cost and is
18.9 unit is spent.As seen from table, given deadline constrain under, scheduling evolution optimization method ratio PSO of the invention with
CGA2It is lower to execute cost for dispatching method constraint satisfaction rate higher.
The Comparative result of table 1 the present embodiment method and other methods
Above-mentioned workflow schedule evolution optimization method the characteristics of for workflow task set, is proposed based on genetic algorithm
One kind evolving (Re-Evolution) strategy to solve the problems, such as that local optimum and convergence rate are slow again, and proposes one kind of multiple dyes
Colour solid can fully describe workflow schedule strategy, and then by adaptive to coding structure according to workflow task nodal information
The lower correlated variables for correcting mean value of deadline constraint is calculated to ensure the survival rate of effective gene, optimizing resources are dispatched from two
According to the preferably final scheduling strategy of established rule in pond.It is compared with the existing dispatching method based on evolution algorithm, this method is reduced
Calculation amount can guarantee the Service Efficiency that specific application context restrictions need, and ensure the controllable execution efficiency of algorithm, can avoid office
The problems such as portion is optimal and convergence rate is slow.Verified, the workflow schedule evolution under the deadline constraint that the present invention announces is sought
Excellent method can meet the actual demand of specified application, have widely in the workflow schedule optimizing field constrained based on deadline
Reference value.
Corresponding to the workflow schedule evolution optimization method described in foregoing embodiments, Fig. 4 shows that the embodiment of the present invention carries
The running environment schematic diagram of the workflow schedule evolution optimizing program of confession.For convenience of description, it illustrates only and the present embodiment phase
The part of pass.
In the present embodiment, the workflow schedule evolution optimizing program 400 is installed and is run in terminal device 40.
The terminal device 40 may include, but be not limited only to, memory 401 and processor 402.Fig. 4 is illustrated only with component 401-
402 terminal device 40, it should be understood that being not required for implementing all components shown, the implementation that can be substituted is more
Or less component.
The memory 401 can be the internal storage unit of the terminal device 40 in some embodiments, such as should
The hard disk or memory of terminal device 40.The memory 401 can also be the terminal device 40 in further embodiments
The plug-in type hard disk being equipped on External memory equipment, such as the terminal device 40, intelligent memory card (Smart Media
Card, SMC), secure digital (Secure Digital, SD) card, flash card (Flash Card) etc..Further, described to deposit
Reservoir 401 can also both include the terminal device 40 internal storage unit and also including External memory equipment.The memory
401 are used to store the application software and Various types of data for being installed on the terminal device 40, such as workflow schedule evolution is sought
The program code etc. of excellent program 400.The memory 401, which can be also used for temporarily storing, have been exported or will export
Data.
The processor 402 can be a central processing unit (Central Processing in some embodiments
Unit, CPU), microprocessor or other data processing chips, for run the program code stored in the memory 401 or
Handle data, such as execute the workflow schedule evolution optimizing program 400 etc..
The terminal device 40 may also include display, the display can be in some embodiments LED displays,
Liquid crystal display, touch-control liquid crystal display and Organic Light Emitting Diode (Organic Light-Emitting Diode,
OLED device etc.) is touched.
Referring to Fig. 5, being the Program modual graph of workflow schedule evolution optimizing program 400 provided in an embodiment of the present invention.
In the present embodiment, the workflow schedule evolution optimizing program 400 can be divided into one or more modules, and described one
A or multiple modules are stored in the memory 401, and (the present embodiment is the processing by one or more processors
Device 402) it is performed, to complete the present invention.For example, in Figure 5, the workflow schedule evolution optimizing program 400 can be by
It is divided into information coding structure design module 501, balanced mixing initialization population to form module 502, scheduling optimizing policy module
503 and determination module 504.The so-called module of the present invention is the series of computation machine program instruction for referring to complete specific function
Section, than program more suitable for describing implementation procedure of the workflow schedule evolution optimizing program 400 in the server 40.
The function of the module 501-504 will specifically be introduced by being described below.
Wherein, information coding structure design module 501, for according to user's input parameter and Work stream data, choosing and losing
Chromosome is as scheduling strategy information carrier in propagation algorithm population, in conjunction with population scale M, computing resource scale N and workflow
Information coding structure of the design data multiple chromosome to W；User's input parameter includes deadline constrained parameters.
Equilibrium mixing initialization population forms module 502, any multiple for being constructed according to described information coding structure
Chromosome pair recycles single a variety of chromosomes to construction process according to building method, it is initial to form the balanced mixing that scale is M
Change population.
Optimizing policy module 503 is dispatched, is used for according to the balanced mixing initialization population, structure evolutionary process, and according to
Evolutionary direction is adjusted according to user's input parameter, mean value construction is corrected according to user's input parameter and original fitness
Fitness is improved, and according to the adaptive crossover mutation Pc of every chromosome of population in fitness calculating evolutionary process and certainly
Every chromosome is carried out genetic manipulation and gained scheduling strategy is stored in scheduling optimizing strategy pond A by adequate variation probability P m；With
And
The input of evolutional operation again under ω chromosome in the A of preference policy pond is constrained as deadline, and be added
The chromosome that M- ω items generate at random obtains again Advanced group species and is executed again as the input of step S3 and according to established rule iteration
It evolves, gained scheduling strategy is stored in scheduling optimizing strategy pond B；
Determination module 504, every scheduling plan for overall merit scheduling optimizing strategy pond A and scheduling optimizing strategy pond B
Slightly, deadline constraints is matched, and compares deadline constraint satisfaction rate and executes cost, evaluation algorithms execution efficiency,
Finally judge preferred scheduling strategy.
As a kind of embodiment, multiple chromosome includes 2 chromosomes altogether to the information coding structure of W, i.e.,
Multiple chromosome is to W={ chromesome1, chromesome2 }
Chromesome1=sequence of { Pc, Pm, gen0,gen1,…,genk,…,genM-1}
Chromesome2={ tof0,tof1,…,tofi,…,tofN-1}
Assuming that on i-th virtual machine, the scale for executing sequence is g, then tofiIt is represented by：
Wherein, Pc is the adaptive crossover mutation of chromosome；Pm is the self-adaptive mutation of chromosome；genkFor dyeing
Gene on the k location of body, 0≤k<M；vmiFor No. i-th virtual machine, 0≤i in computing resource<N； tofiMoney is calculated to be assigned to
Source vmiCalculating task execute sequence；Gene gen's is encoded to t (vm), expression calculating task in chromesome1 chromosomes
T distributes computing resource vm；Gene vm indicates that computing resource, subsequent gene set { t } indicate to divide in chromesome2 chromosomes
Fit over the calculating task of computing resource vm.
As another embodiment, equilibrium mixing initialization population forms module 502 and is specifically used for：
The chromosome obtained according to step S1 is to coding structure, it is assumed that use k kind method construct initialization populations, then according to
Method 1 constructs x1Chromosome pair constructs x according to method iiChromosome pair constructs x according to method kkChromosome pair, obtains
The balanced mixing initial population for being M to scale；Wherein, k >=2,
Optionally, scheduling optimizing policy module 503 corrects mean value according to user's input parameter and original fitness
Construction improves fitness, and the adaptive crossover mutation Pc of every chromosome of population in evolutionary process is calculated according to the fitness
With self-adaptive mutation Pm, detailed process is as follows：
For the balanced mixing initialization population, the original fitness vector for calculating chromosome is constrained according to deadline
Fini, mean value is corrected by original fitness, construction improves fitness vector Fimp；
Using greatest improvement fitness value subtract it is current improve fitness value, and divided by greatest improvement fitness value with improve
Fitness corrects the difference of mean value, obtains probability intermediate variable Vmi；
When being less than its amendment mean value to improving fitness value, it is 1 that probability, which corrects variable-value,；
The relationship of mean value is corrected with it to improving fitness value, is multiplied respectively with different parameters, is labeled as Pc；
M target chromosome sample is taken, above procedure is repeated and obtains M Pc value；
Based on same process, Pm can be obtained.
Further, determination module 504 is specifically used for：
Step S2, step S3 are executed to all chromosomes in population and step S4 obtains scheduling optimizing strategy pond A and tune
Optimizing strategy pond B is spent, using deadline constraint satisfaction rate as evaluation criterion, compares the obtained scheduling optimizing of step S4 and step S5
Tactful pond A and scheduling optimizing strategy pond B；
According to user's input information, secondary constraint is determined；User's input information further includes secondary constraint information；
If a certain scheduling optimizing strategy fitness is optimal and does not violate secondary deadline constraint, or violates deadline about
Shu Chengdu is minimum with respect to other scheduling optimizing strategies, then judges the scheduling optimizing strategy as target dispatch optimizing strategy, and give
Go out scheduling result.
It is apparent to those skilled in the art that for convenience of description and succinctly, only with above-mentioned each work(
Can unit, module division progress for example, in practical application, can be as needed and by above-mentioned function distribution by different
Functional unit, module are completed, i.e., the internal structure of described device are divided into different functional units or module, more than completion
The all or part of function of description.Each functional unit, module in embodiment can be integrated in a processing unit, also may be used
It, can also be above-mentioned integrated during two or more units are integrated in one unit to be that each unit physically exists alone
The form that hardware had both may be used in unit is realized, can also be realized in the form of SFU software functional unit.In addition, each function list
Member, the specific name of module are also only to facilitate mutually distinguish, the protection domain being not intended to limit this application.Above system
The specific work process of middle unit, module, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
Those of ordinary skill in the art may realize that lists described in conjunction with the examples disclosed in the embodiments of the present disclosure
Member and algorithm steps can be realized with the combination of electronic hardware or computer software and electronic hardware.These functions are actually
It is implemented in hardware or software, depends on the specific application and design constraint of technical solution.Professional technician
Each specific application can be used different methods to achieve the described function, but this realization is it is not considered that exceed
The scope of the present invention.
In embodiment provided by the present invention, it should be understood that disclosed device and method can pass through others
Mode is realized.For example, system embodiment described above is only schematical, for example, the division of the module or unit,
Only a kind of division of logic function, formula that in actual implementation, there may be another division manner, such as multiple units or component can be with
In conjunction with or be desirably integrated into another system, or some features can be ignored or not executed.Another point, it is shown or discussed
Mutual coupling or direct-coupling or communication connection can be by some interfaces, the INDIRECT COUPLING of device or unit or
Communication connection can be electrical, machinery or other forms.
The unit illustrated as separating component may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, you can be located at a place, or may be distributed over multiple
In network element.Some or all of unit therein can be selected according to the actual needs to realize the mesh of this embodiment scheme
's.
In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, it can also
It is that each unit physically exists alone, it can also be during two or more units be integrated in one unit.Above-mentioned integrated list
The form that hardware had both may be used in member is realized, can also be realized in the form of SFU software functional unit.
If the integrated unit is realized in the form of SFU software functional unit and sells or use as independent product
When, it can be stored in a computer read/write memory medium.Based on this understanding, the technical solution of the embodiment of the present invention
Substantially all or part of the part that contributes to existing technology or the technical solution can be with software product in other words
Form embody, which is stored in a storage medium, including some instructions use so that one
Computer equipment (can be personal computer, server or the network equipment etc.) or processor (processor) execute this hair
The all or part of step of bright each embodiment the method.And storage medium above-mentioned includes：USB flash disk, read-only is deposited mobile hard disk
Reservoir (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disc or
The various media that can store program code such as CD.
Embodiment described above is merely illustrative of the technical solution of the present invention, rather than its limitations；Although with reference to aforementioned reality
Applying example, invention is explained in detail, it will be understood by those of ordinary skill in the art that：It still can be to aforementioned each
Technical solution recorded in embodiment is modified or equivalent replacement of some of the technical features；And these are changed
Or replace, the spirit and scope for various embodiments of the present invention technical solution that it does not separate the essence of the corresponding technical solution should all
It is included within protection scope of the present invention.
Claims (10)
1. a kind of workflow schedule evolution optimization method, which is characterized in that including：
Step S1 chooses in Population in Genetic Algorithms chromosome as scheduling strategy according to user's input parameter and Work stream data
Information carrier designs information coding of the multiple chromosome to W in conjunction with population scale M, computing resource scale N and Work stream data
Structure；User's input parameter includes deadline constrained parameters；
Step S2 is single more according to building method cycle according to any multiple chromosome pair that described information coding structure constructs
Kind chromosome forms the balanced mixing initialization population that scale is M to construction process；
Step S3 builds evolutionary process, and adjust according to user's input parameter according to the balanced mixing initialization population
Evolutionary direction corrects mean value construction according to user's input parameter and original fitness and improves fitness, and according to described
Fitness calculates the adaptive crossover mutation Pc and self-adaptive mutation Pm of every chromosome of population in evolutionary process, by every
Chromosome carries out genetic manipulation and gained scheduling strategy is stored in scheduling optimizing strategy pond A；
Step S4, the input of evolutional operation again under ω chromosome in the A of preference policy pond is constrained as deadline, and add
Enter the chromosome that M- ω items generate at random, obtains again Advanced group species and executed as the input of step S3 and according to established rule iteration
It evolves again, gained scheduling strategy is stored in scheduling optimizing strategy pond B；
Step S5, overall merit dispatches optimizing strategy pond A and dispatches every scheduling strategy of optimizing strategy pond B, when matching ends
Between constraints, and compare deadline constraint satisfaction rate and execute costs, evaluation algorithms execution efficiency, final judgement preferably tune
Degree strategy.
2. workflow schedule evolution optimization method according to claim 1, which is characterized in that the multiple staining in step S1
Body includes 2 chromosomes altogether to the information coding structure of W, i.e.,
Multiple chromosome is to W={ chromesome1, chromesome2 }
Chromesome1=sequence of { Pc, Pm, gen0,gen1,…,genk,…,genM-1}
Chromesome2={ tof0,tof1,…,tofi,…,tofN-1}
Assuming that on i-th virtual machine, the scale for executing sequence is g, then tofiIt is represented by：
Wherein, Pc is the adaptive crossover mutation of chromosome；Pm is the self-adaptive mutation of chromosome；genkFor the k of chromosome
Gene on position, 0≤k<M；vmiFor No. i-th virtual machine, 0≤i in computing resource<N；tofiTo be assigned to computing resource vmi
Calculating task execute sequence；Gene gen's is encoded to t (vm), the t distribution of expression calculating task in chromesome1 chromosomes
Computing resource vm；Gene vm indicates that computing resource, subsequent gene set { t } indicate distribution at this in chromesome2 chromosomes
The calculating task of computing resource vm.
3. workflow schedule evolution optimization method according to claim 1, which is characterized in that encoded and tied according to described information
Any multiple chromosome pair of structure construction recycles single a variety of chromosomes to construction process according to building method, and it is M to form scale
Balanced mixing initialization population, detailed process is as follows：
The chromosome obtained according to step S1 is to coding structure, it is assumed that uses k kind method construct initialization populations, then according to method
1 construction x1Chromosome pair constructs x according to method iiChromosome pair constructs x according to method kkChromosome pair, obtains scale
For the balanced mixing initial population of M；Wherein, k >=2,
4. workflow schedule evolution optimization method according to claim 1, which is characterized in that input and join according to the user
Several and original fitness corrects mean value construction and improves fitness, and calculates population every in evolutionary process according to the fitness
The adaptive crossover mutation Pc and self-adaptive mutation Pm, detailed process of chromosome are as follows：
For the balanced mixing initialization population, the original fitness vector F for calculating chromosome is constrained according to deadlineini,
Mean value is corrected by original fitness, construction improves fitness vector Fimp；
Using greatest improvement fitness value subtract it is current improve fitness value, and divided by greatest improvement fitness value adapt to improving
Degree corrects the difference of mean value, obtains probability intermediate variable Vmi；
When being less than its amendment mean value to improving fitness value, it is 1 that probability, which corrects variable-value,；
The relationship of mean value is corrected with it to improving fitness value, is multiplied respectively with different parameters, is labeled as Pc；
M target chromosome sample is taken, above procedure is repeated and obtains M Pc value；
Based on same process, Pm can be obtained.
5. workflow schedule evolution optimization method according to claim 1, which is characterized in that judge preferred adjust in step S5
The process for spending strategy is specific as follows：
Step S2, step S3 are executed to all chromosomes in population and step S4 obtain scheduling optimizing strategy pond A and scheduling is sought
Dominant strategy pond B, using deadline constraint satisfaction rate as evaluation criterion, comparison step S4 and step S5 obtains scheduling optimizing strategy
Pond A and scheduling optimizing strategy pond B；
According to user's input information, secondary constraint is determined；User's input information further includes secondary constraint information；
If a certain scheduling optimizing strategy fitness is optimal and does not violate secondary deadline constraint, or violates deadline constraint journey
Other opposite scheduling optimizing strategies of degree are minimum, then judge the scheduling optimizing strategy as target dispatch optimizing strategy, and provide tune
Spend result.
6. a kind of terminal device, which is characterized in that including memory, processor, being stored in the memory can be at the place
The computer program run on reason device, the processor realize following steps when executing the computer program：
Step S1 chooses in Population in Genetic Algorithms chromosome as scheduling strategy according to user's input parameter and Work stream data
Information carrier designs information coding of the multiple chromosome to W in conjunction with population scale M, computing resource scale N and Work stream data
Structure；User's input parameter includes deadline constrained parameters；
Step S2 is single more according to building method cycle according to any multiple chromosome pair that described information coding structure constructs
Kind chromosome forms the balanced mixing initialization population that scale is M to construction process；
Step S3 builds evolutionary process, and adjust according to user's input parameter according to the balanced mixing initialization population
Evolutionary direction corrects mean value construction according to user's input parameter and original fitness and improves fitness, and according to described
Fitness calculates the adaptive crossover mutation Pc and self-adaptive mutation Pm of every chromosome of population in evolutionary process, by every
Chromosome carries out genetic manipulation and gained scheduling strategy is stored in scheduling optimizing strategy pond A；
Step S4, the input of evolutional operation again under ω chromosome in the A of preference policy pond is constrained as deadline, and add
Enter the chromosome that M- ω items generate at random, obtains again Advanced group species and executed as the input of step S3 and according to established rule iteration
It evolves again, gained scheduling strategy is stored in scheduling optimizing strategy pond B；
Step S5, overall merit dispatches optimizing strategy pond A and dispatches every scheduling strategy of optimizing strategy pond B, when matching ends
Between constraints, and compare deadline constraint satisfaction rate and execute costs, evaluation algorithms execution efficiency, final judgement preferably tune
Degree strategy.
7. terminal device according to claim 6, which is characterized in that the multiple chromosome in step S1 compiles the information of W
Code structure includes 2 chromosomes altogether, i.e.,
Multiple chromosome is to W={ chromesome1, chromesome2 }
Chromesome1=sequence of { Pc, Pm, gen0,gen1,…,genk,…,genM-1}
Chromesome2={ tof0,tof1,…,tofi,…,tofN-1}
Assuming that on i-th virtual machine, the scale for executing sequence is g, then tofiIt is represented by：
Wherein, Pc is the adaptive crossover mutation of chromosome；Pm is the self-adaptive mutation of chromosome；genkFor the k of chromosome
Gene on position, 0≤k<M；vmiFor No. i-th virtual machine, 0≤i in computing resource<N；Tofi is to be assigned to computing resource vmi
Calculating task execute sequence；Gene gen's is encoded to t (vm), the t distribution of expression calculating task in chromesome1 chromosomes
Computing resource vm；Gene vm indicates that computing resource, subsequent gene set { t } indicate distribution at this in chromesome2 chromosomes
The calculating task of computing resource vm.
8. terminal device according to claim 6, which is characterized in that according to any more of described information coding structure construction
Colour solid pair is redyed, single a variety of chromosomes is recycled to construction process according to building method, it is initial to form the balanced mixing that scale is M
Change population, detailed process is as follows：
The chromosome obtained according to step S1 is to coding structure, it is assumed that uses k kind method construct initialization populations, then according to method
1 construction x1Chromosome pair constructs x according to method iiChromosome pair constructs x according to method kkChromosome pair, obtains scale
For the balanced mixing initial population of M；Wherein, k >=2,
9. terminal device according to claim 6, which is characterized in that according to user's input parameter and original adaptation
Degree correct mean value construction improve fitness, and according to the fitness calculate evolutionary process in every chromosome of population it is adaptive
Crossover probability Pc and self-adaptive mutation Pm, detailed process are as follows：
For the balanced mixing initialization population, the original fitness vector F for calculating chromosome is constrained according to deadlineini,
Mean value is corrected by original fitness, construction improves fitness vector Fimp；
Using greatest improvement fitness value subtract it is current improve fitness value, and divided by greatest improvement fitness value adapt to improving
Degree corrects the difference of mean value, obtains probability intermediate variable Vmi；
When being less than its amendment mean value to improving fitness value, it is 1 that probability, which corrects variable-value,；
The relationship of mean value is corrected with it to improving fitness value, is multiplied respectively with different parameters, is labeled as Pc；
M target chromosome sample is taken, above procedure is repeated and obtains M Pc value；
Based on same process, Pm can be obtained.
10. a kind of computer readable storage medium, the computer-readable recording medium storage has computer program, feature to exist
In when the computer program is executed by processor the step of any one of such as claim 1 to 5 of realization the method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810154127.7A CN108320059B (en) | 2018-02-22 | 2018-02-22 | Workflow scheduling evolution optimization method and terminal equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810154127.7A CN108320059B (en) | 2018-02-22 | 2018-02-22 | Workflow scheduling evolution optimization method and terminal equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108320059A true CN108320059A (en) | 2018-07-24 |
| CN108320059B CN108320059B (en) | 2020-06-09 |
Family
ID=62900738
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810154127.7A Active CN108320059B (en) | 2018-02-22 | 2018-02-22 | Workflow scheduling evolution optimization method and terminal equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108320059B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110516795A (en) * | 2019-08-28 | 2019-11-29 | 北京达佳互联信息技术有限公司 | A kind of method, apparatus and electronic equipment for model variable allocation processing device |
| CN111382941A (en) * | 2020-03-09 | 2020-07-07 | 河海大学常州校区 | Parallel task scheduling method with multiple constraint conditions |
| CN111882234A (en) * | 2020-08-03 | 2020-11-03 | 浪潮云信息技术股份公司 | Scientific workflow task management method and device |
| CN112256925A (en) * | 2020-10-21 | 2021-01-22 | 西安电子科技大学 | Multi-request-oriented scientific workflow data set storage method |
| CN112884383A (en) * | 2021-04-19 | 2021-06-01 | 上海海事大学 | Container port emergency material optimizing and dispatching method considering time window constraint |
| CN113434267A (en) * | 2021-05-25 | 2021-09-24 | 深圳大学 | Cloud computing workflow dynamic scheduling method, device, equipment and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107133091A (en) * | 2017-05-08 | 2017-09-05 | 武汉轻工大学 | The cloud workflow task dispatching method being classified based on top-down task |
| CN107329831A (en) * | 2017-06-29 | 2017-11-07 | 北京仿真中心 | A kind of artificial resource dispatching method based on improved adaptive GA-IAGA |
-
2018
- 2018-02-22 CN CN201810154127.7A patent/CN108320059B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107133091A (en) * | 2017-05-08 | 2017-09-05 | 武汉轻工大学 | The cloud workflow task dispatching method being classified based on top-down task |
| CN107329831A (en) * | 2017-06-29 | 2017-11-07 | 北京仿真中心 | A kind of artificial resource dispatching method based on improved adaptive GA-IAGA |
Non-Patent Citations (4)
| Title |
|---|
| JAILALITA: "《Critical path based scheduling algorithm for workflow applications in cloud computing》", 《2016 ICACCA（SPRING）》 * |
| V.VINOTHINA: "《Scheduling Scientific Workflow Tasks in Cloud Using Swarm Intelligence》", 《2017 IEEE INTERNATIONAL CONFERENCE ON CURRENT TRENDS IN ADVANCED COMPUTING (ICCTAC)》 * |
| 党云龙: "《截止时间约束的工作流调度自适应进化方法》", 《石家庄铁道大学学报》 * |
| 梁合兰: "《时序约束下科学工作流的动态调度研究》", 《系统工程理论与实践》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110516795A (en) * | 2019-08-28 | 2019-11-29 | 北京达佳互联信息技术有限公司 | A kind of method, apparatus and electronic equipment for model variable allocation processing device |
| CN110516795B (en) * | 2019-08-28 | 2022-05-10 | 北京达佳互联信息技术有限公司 | Method and device for allocating processors to model variables and electronic equipment |
| CN111382941A (en) * | 2020-03-09 | 2020-07-07 | 河海大学常州校区 | Parallel task scheduling method with multiple constraint conditions |
| CN111382941B (en) * | 2020-03-09 | 2022-08-02 | 河海大学常州校区 | Parallel task scheduling method with multiple constraint conditions |
| CN111882234A (en) * | 2020-08-03 | 2020-11-03 | 浪潮云信息技术股份公司 | Scientific workflow task management method and device |
| CN112256925A (en) * | 2020-10-21 | 2021-01-22 | 西安电子科技大学 | Multi-request-oriented scientific workflow data set storage method |
| CN112884383A (en) * | 2021-04-19 | 2021-06-01 | 上海海事大学 | Container port emergency material optimizing and dispatching method considering time window constraint |
| CN112884383B (en) * | 2021-04-19 | 2024-04-05 | 上海海事大学 | Container port emergency material optimizing and transferring method considering time window constraint |
| CN113434267A (en) * | 2021-05-25 | 2021-09-24 | 深圳大学 | Cloud computing workflow dynamic scheduling method, device, equipment and storage medium |
| CN113434267B (en) * | 2021-05-25 | 2022-12-02 | 深圳大学 | Cloud computing workflow dynamic scheduling method, device, equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108320059B (en) | 2020-06-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108320059A (en) | A kind of workflow schedule evolution optimization method and terminal device | |
| Alkayal et al. | Efficient task scheduling multi-objective particle swarm optimization in cloud computing | |
| Zhu et al. | An efficient evolutionary grey wolf optimizer for multi-objective flexible job shop scheduling problem with hierarchical job precedence constraints | |
| Akbari et al. | An enhanced genetic algorithm with new operators for task scheduling in heterogeneous computing systems | |
| CN110489229B (en) | Multi-target task scheduling method and system | |
| CN113411369B (en) | Cloud service resource collaborative optimization scheduling method, system, medium and equipment | |
| Gao et al. | Pareto-based grouping discrete harmony search algorithm for multi-objective flexible job shop scheduling | |
| Ashouraie et al. | Priority-based task scheduling on heterogeneous resources in the Expert Cloud | |
| CN106951330A (en) | A kind of maximized virtual machine distribution method of cloud service center service utility | |
| Gao et al. | A hybrid algorithm for multi-objective scientific workflow scheduling in IaaS cloud | |
| Xiong et al. | A Hybrid Multiobjective Evolutionary Approach for Flexible Job‐Shop Scheduling Problems | |
| Durgadevi et al. | Resource allocation in cloud computing using SFLA and cuckoo search hybridization | |
| CN111813500B (en) | Multi-target cloud workflow scheduling method and device | |
| CN110008023B (en) | Cloud computing system budget constraint random task scheduling method based on genetic algorithm | |
| Kalra et al. | Multi‐criteria workflow scheduling on clouds under deadline and budget constraints | |
| CN109255513A (en) | A kind of power telecom network scene work order dispatching method | |
| Patra et al. | Game theoretic approach for real-time task scheduling in cloud computing environment | |
| Peng et al. | Genetic Algorithm‐Based Task Scheduling in Cloud Computing Using MapReduce Framework | |
| Moazeni et al. | Dynamic resource allocation using an adaptive multi-objective teaching-learning based optimization algorithm in cloud | |
| Zhang et al. | Individualized requirement-driven multi-task scheduling in cloud manufacturing using an extended multifactorial evolutionary algorithm | |
| Shafahi et al. | An efficient task scheduling in cloud computing based on ACO algorithm | |
| CN112884248B (en) | Optimization method of large-scale cloud service process | |
| Yang et al. | A fully hybrid algorithm for deadline constrained workflow scheduling in clouds | |
| Nguyen et al. | A novel nature-inspired algorithm for optimal task scheduling in fog-cloud blockchain system | |
| Ananth et al. | Cooperative game theoretic approach for job 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |