CN104331321B - Cloud computing task scheduling method based on tabu search and load balancing - Google Patents
Cloud computing task scheduling method based on tabu search and load balancing Download PDFInfo
- Publication number
- CN104331321B CN104331321B CN201410527189.XA CN201410527189A CN104331321B CN 104331321 B CN104331321 B CN 104331321B CN 201410527189 A CN201410527189 A CN 201410527189A CN 104331321 B CN104331321 B CN 104331321B
- Authority
- CN
- China
- Prior art keywords
- task
- virtual machine
- individual
- scheduling
- load
- 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.)
- Expired - Fee Related
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a cloud computing task scheduling method based on tabu search and load balancing. According to the method, a task scheduling initial solution is obtained on the basis of a heuristic priority allocation strategy of the earliest finish time; then, candidate exchange task pairs are generated on the basis of using the tabu search as an instructional neighborhood search optimization strategy, the task pairs with great profit values are exchanged by adopting a greedy principle, and the task scheduling initial solution is optimized, so the completion time span of the whole task is shortened to the greatest degree. After the cloud computing task scheduling method based on tabu search and load balancing is adopted, the task scheduling efficiency is effectively improved, the load balancing of a cloud computing platform is realized, the idle time of resources is also obviously reduced, the utilization benefits of the resources are improved, and better practicability is realized.
Description
Technical field
The present invention relates under a kind of cloud computing environment based on TABU search and the cloud computing task scheduling side of load balancing
Method.
Background technology
Cloud computing is used as the conventional arts such as Distributed Calculation, parallel computation, grid computing and network programming model, distributed
The product of the new technique fusion development such as data storage technology, Intel Virtualization Technology, is the key strategy for leading information industry to innovate
Property technology and means, have important strategic importance to China's developing new and high-tech industry.Cloud computing is by the way that calculating task is drawn
Divide on large-scale low-cost server cluster so that people can process more multiple using the unused resource for being distributed in various places
Miscellaneous application program, with extremely low cost input high calculating quality is obtained.
The essence of task scheduling is the thing that n separate task is assigned to m unused isomery under cloud computing environment
In reason resource so that the general assignment deadline is minimum and available resources are fully used.Task scheduling is flat as cloud computing
The important component part of platform, its efficiency directly influences the performance and service quality of whole cloud computing platform.For example, serial task
Dispatching method is sequentially assigned to one group of virtual machine a group task, the task of each virtual machine operation equal number is ensured as far as possible with
Balanced load, but do not account for the difference between the demand of task and virtual machine.Mission Scheduling has proven to one
Np complete problem, in mnThe solution space of individual possible task scheduling finds approximate optimal solution so that execution time of general assignment and negative
Equilibrium degree is carried minimum, wherein it is that, in order to improve service quality, load balancing degrees minimum is to ensure cloud ring to perform time minimum
The stability in border.
Task scheduling refers to the scheduling of Meta task in cloud computing environment according to the present invention, i.e., separate between task,
Its scheduling does not consider the data correlation and priority constraint relationship between task.At present, the method for scheduling task of cloud computing has not yet been formed
Unified standards and norms, but due to the importance of the problem, domestic and international researcher proposes substantial amounts of cloud computing task scheduling
Method comes the approximate optimal solution of calculating task scheduling, Min-Min, Max-Min, the Dynamic Programming in existing conventional mesh calculating etc.
Heuristic mutation operations method, also has based on genetic algorithm, particle cluster algorithm, ant group algorithm, immune algorithm and differential evolution algorithm etc.
Intelligent dispatching method.Wherein, the Min-Min algorithms of traditional heuristic mutation operations method adopt easy first and difficult later strategy, have first carried out
Into the task that the time is short, the task of deadline length is then performed, and takes Greedy strategy that each priority of task is assigned to and held
The capable computing resource that it is completed earliest；Max-Min algorithms are on the contrary, using difficult at first and quite easy afterwards and Greedy strategy, choose every time
Deadline most long task, is preferentially assigned to and performs the computing resource that it is completed earliest.Traditional heuristic mutation operations method is with most
The early deadline is scheduled for target, there is preferable load-balancing performance, but the actual execution time of general assignment is not most
It is few.Intelligent dispatching method by the coding to task scheduling approach, and according to genetic algorithm, particle cluster algorithm, ant group algorithm, exempt from
The intelligent algorithm thought such as epidemic disease algorithm and differential evolution algorithm, in mnBetween the solution space Diversification of size and Intensification
Equilibrium establishment, finally significantly reduces the execution time of task.However, intelligent dispatching method is carrying out magnanimity task scheduling mistake
Cheng Zhong, is easily absorbed in locally optimal solution, and the effect in terms of convergence rate and load balancing has much room for improvement.
The content of the invention
Present invention aims to the deficiency that prior art is present, there is provided one kind is based on TABU search and load balancing
Cloud computing method for scheduling task, solve the optimization problem for performing time and load balancing under cloud computing environment in task scheduling,
The time span that the task of being effectively shortened is completed, realizes the reasonable utilization of cloud computing resources, provides efficiently for cloud computing
Task Scheduling Mechanism.To reach above-mentioned purpose, the design of the present invention is as follows.
First, on the basis of the formalized description of the Load Balancing Task Scheduling problem under cloud computing environment, by dynamic rule
The formalizing deduction of the method for drawing obtains the heuristic preferential allocation strategy on earliest finish time, and is tried to achieve based on the allocation strategy and appoint
The initial solution of business scheduling.
2nd, by the financial value concept of task switching, based on TABU search as guiding neighborhood search optimisation strategy come
Candidate's switching task pair is produced, selects the big task of financial value to swapping using greedy principle, optimize the first of task scheduling
Begin solution, so as to farthest shorten the time span that whole task is completed, and realizes the load balancing of cloud computing platform.Task
The initial solution of scheduling optimizes the key link as cloud computing Task Scheduling Mechanism, operation of its result to whole cloud computing environment
Efficiency has important impact, can efficiently reduce resource free time, improves the utilization benefit of resource.
According to above-mentioned inventive concept, the technical scheme is that what is be achieved in that：One kind is based on multilevel partitioning
With the cloud computing method for scheduling task for assigning power Directed Hypergraph, it is characterised in that comprise the following steps that.
Step 1, class types degree analysis, is input into the task that user submits under cloud computing environment, and carries out type and class to it
The analysis of degree, determines the parallelization degree and feature of task.
Step 2, proceeding graininess decomposes, and is total to according to the parallelization degree and feature of user task, and the resource of cloud computing
Method of salary distribution peculiar property is enjoyed, user task is decomposed according to proceeding graininess rank.
Step 3, resource characteristicses analysis, according to the resource-sharing method of salary distribution peculiar property of cloud computing, to decomposition after appoint
Business carries out resource characteristicses analysis.
Step 4, solves task scheduling initial solution, according to the analysis result to task resource characteristic, sets up and describes its resource
Demand model, and then the initial solution of task scheduling is tried to achieve based on the model.
Step 5, optimizes task scheduling initial solution, optimizes task scheduling initial solution, shortens the latest finishing time of general assignment
And improve the load-balancing performance of virtual machine, obtain the optimization solution of task scheduling.
Step 6, duty mapping scheduling, by MapReduce Task Scheduling Models, reflects to the optimization solution of task scheduling
Penetrate and dispatch, the task in cloud computing environment of realizing is submitted to and performed, effectively the load and shortening of balanced cloud computing platform
The time span that whole task is completed.
It is as follows the step of described solution task scheduling initial solution in above-mentioned step 4.
Step 4.1, the assignment instructions length and virtual machine execute instruction bar number per second be given based on Resources requirement model, meter
Calculate set of tasksTInnIndividual task is in virtual machine setVM'smThe expected execution time on individual virtual machine, obtainn×mExpection
Perform time matrixC,Wherein it is expected execution time CijRepresent theiIndividual task isjThe time performed on individual virtual machine, equal to theiThe command length of individual task is divided byjThe execute instruction bar number per second of individual virtual machine.
Step 4.2, initializes the present load array of m virtual machinevt[1..m] be zero, i.e., start distribution task it
The present load of front any virtual machine is zero.
Step 4.3, sequential access set of tasksTIn each task, heuristic preferential point based on earliest finish time
With strategy, successively bykIndividual task is distributed on the virtual machine with earliest finish time, until after the distribution of all tasks terminates
Obtain the initial solution of task scheduling.
It is as follows the step of described optimization task scheduling initial solution in above-mentioned step 5.
Step 5.1, initialization task setTInnThe taboo array of individual tasktabu[1..n] be zero, that is, allow all of
Task is exchanged.
Step 5.2, selects the big task of financial value to swapping, until cannot subtract based on TABU search and greedy principle
The latest finishing time of few general assignment.
In above-mentioned step 4.3, the described heuristic preferential allocation strategy based on earliest finish time is bykIndividual task
Distribute to step on the virtual machine with earliest finish time as follows.
Step 4.3.1, foundationmThe present load array of individual virtual machinevt[1..m] and expected execution time matrixC, calculate
Go outkIndividual taskt k Distribute time span corresponding to each virtual machinemakespan, wherein thejThe time span of individual virtual machine
ForjThe present load array of individual virtual machinevt[j] and thekIndividual taskt k jThe execution time of individual virtual machinec kj Sum.
Step 4.3.2, finds out the minimum virtual machine of time spanvm x 。
Step 4.3.3, distributes taskt k To virtual machinevm x , updatevm x Virtual machine is loadedvt[x] for thexIndividual virtual machine
Present load arrayvt[x] and thekIndividual taskt k xThe execution time of individual virtual machinec kx Sum.
In above-mentioned step 5.2, described selects the big task of financial value to carrying out based on TABU search and greedy principle
Exchange step is as follows.
Step 5.2.1, is based onmThe load array of individual virtual machinevt[1..m], find out the maximum virtual machine of loadvm x With
The minimum virtual machine of loadvm y 。
Step 5.2.2, if distributed to the maximum virtual machine of loadvm x All tasks be prohibited exchange, then task hand over
The optimization solution for terminating and obtaining task scheduling is changed, 6 are gone to step；Otherwise in virtual machinevm x Select the allowing to be exchanged of the taskt k , and by the taskt k Corresponding taboo marktabu[kIt is revised as]=0tabu[k]=1。
Step 5.2.3, if distributed to the minimum virtual machine of loadvm y All of task is prohibited to exchange, then task is handed over
The optimization solution for terminating and obtaining task scheduling is changed, 6 are gone to step；Otherwise in virtual machinevm y Calculate it is all allow exchanged appoint
Business and taskt k Financial value after exchange.If the exchange financial value of all tasks pair for calculating is negative, step is skipped
5.2.4,5.2.5,5.2.6 and 5.2.7, continue cycling through execution step 5.2.1, otherwise execution step 5.2.4.
Step 5.2.4, select to exchange the maximum task of financial value to (t l , t k ) exchange, i.e. taskt k Exchanged to virtual
Machinevm y Upper execution and taskt l Exchanged to virtual machinevm x Upper execution.If virtual machine after exchangingvm y New load more than virtual
Machinevm x Former load, skip step 5.2.5,5.2.6 and 5.2.7, continue cycling through execution step 5.2.1, otherwise execution step
5.2.5。
Step 5.2.5, changes taskt l Taboo marktabu[l]=1。
Step 5.2.6, updatesvm x Virtual machine is loadedvt[x]=vt[x]+c lx -c kx 。
Step 5.2.7, updatesvm y Virtual machine is loadedvt[y]=vt[y]+c ky -c ly 。
The present invention compared with prior art, with following substantive distinguishing features and remarkable advantage is obviously projected.
1st, improve the efficiency of task scheduling.
The present invention based on TABU search and the cloud computing method for scheduling task of load balancing, by the income of task switching
Value concept, candidate's switching task pair is produced based on TABU search as guiding neighborhood search optimisation strategy, using greedy former
The big task of financial value is then selected to swapping, mapping and the tune to carrying out task after the initial solution optimization of task scheduling again
Degree, so as to be effectively improved the efficiency of task scheduling, the time span that the task of shortening is completed realizes cloud computing resources
Rationally utilize, for cloud computing efficient Task Scheduling Mechanism is provided.
Description of the drawings
Combined by the following example to the present invention based on TABU search and the cloud computing method for scheduling task of load balancing
The description of its accompanying drawing, it will be further appreciated that the purpose of the present invention, specific structural features and advantage.
Fig. 1 is the flow chart of the cloud computing method for scheduling task based on TABU search and load balancing.
Fig. 2 is the flow chart that the heuristic preferential allocation strategy based on earliest finish time solves task scheduling initial solution.
Fig. 3 is the flow chart for optimizing task scheduling initial solution based on TABU search and greedy principle.
Specific embodiment
In order to be more clearly understood that the present invention based on TABU search and the cloud computing method for scheduling task of load balancing
Technology contents, especially exemplified by following instance describe in detail.
The present embodiment based on TABU search and flow chart such as Fig. 1 institutes of the cloud computing method for scheduling task of load balancing
Show.Under cloud computing environment, the task 101 that user submits to is input into, the analysis 102 of type and class degree is carried out to user task, really
Determine the parallelization degree and feature of task；According to the parallelization degree and feature of user task, and the resource-sharing of cloud computing
The peculiar properties such as the method for salary distribution, decomposition 103 is carried out to user task according to proceeding graininess rank；And then the task after decomposition is entered
Row resource characteristicses analysis 104；According to the analysis result to task resource characteristic, set up and describe its resource requirement, and then based on this
Model tries to achieve the initial solution 105 of task scheduling；Optimization task scheduling initial solution, shortens the latest finishing time of general assignment and improves
The load-balancing performance of virtual machine, obtains the optimization solution 106 of task scheduling；By MapReduce Task Scheduling Models, to task
The optimization solution of scheduling is mapped and is dispatched 107；In cloud computing environment, the task that scheduling is completed is submitted to and 108 are performed, from
And effectively shorten the load of time span that whole task completes and balanced cloud computing platform.
The related definition that disclosure sets forth Load Balancing Task Scheduling problem under cloud computing environment is as follows.
Define 1：Under assuming cloud computing environment, user submits breakdown of operation to into the set of n task, and between task mutually
Independent, its scheduling defines set of tasks without the concern for the data correlation and priority constraint relationship between task,
Wherein tiFor i-th task for resolving into, n is the task quantity after decomposing, and i-th task tiTotal finger
Length is made to be MIi。
Define 2：Under assuming cloud computing environment, task scheduling is participated in the set for having m virtual resource, and virtual resource passes through
Virtual machine mode is provided, i.e., virtual resource is the virtual machine in cloud computing cluster.Define virtual machine set,
Wherein vmjFor j-th resources of virtual machine, m is virtual machine quantity, and j-th virtual machine vmjInstruction hold
Scanning frequency degree (execute instruction bar number per second) is MIPSj。
Define 3：Assume that task quantity n after decomposing is not less than resources of virtual machine quantity m (n >=m), each task can only
Distribute to a virtual machine to perform, and a task can only be performed in one virtual machine of certain time period, it is impossible to while performing many
Individual task.Define the square that expected execution time C of the n different task scheduling to m different virtual machine is a n × m
Battle array, wherein CijRepresent i-th task tiIn j-th virtual machine vmjThe time of upper execution, i.e., it is expected to perform
Time CijFor task tiTotal command length MIiDivided by virtual machine vmjExecute instruction bar number MIPS per secondj。
Define 4：Define n different taskIt is dispatched to m different virtual machineOn
All possible task allocative decision collection is.DefinitionRepresent task allocative decision collectionIn a kind of allocative decision, i.e., one
The matrix of individual n × m.Wherein, xijIt is expressed as task tiWith virtual machine vmjThe relations of distribution, and,,,.If i.e. task tiDistribution is in virtual machine vmjUpper execution, then xij=1, otherwise xij=0。
Define 5：For certain task allocative decision, define the present load of virtual machineIt is (front under for current state
The state that k-1 task is assigned), distribute to j-th virtual machine vmjAll required by task perform the time, i.e.,.Define k-th task tkDistribution is in j-th virtual machine vmjOn time spanFor
Task tkIn vmjThe earliest finish time of upper execution, i.e.,。
Define 6：For certain task allocative decision, define the load of virtual machineTo distribute to j-th virtual machine vmj
The expected deadline of all tasks, i.e.,。
Define 7：The average load that n different task is dispatched on m different virtual machine is defined, it is total equal to n task
Command length divided by m virtual machine instructions perform speed it is cumulative and, i.e. general assignment optimal finish time。
Define 8：For certain task allocative decision, define the load balancing degrees of virtual machine。
Load balancing degreesNumerical value is less, shows that the load in cloud computing system between each virtual machine is more balanced.
Define 9：For n different taskIt is dispatched to m different virtual machine
Mission Scheduling be find allocative decisionSo that the task latest finishing time of virtual machine is earliest in the allocative decision, the most long process time of each virtual machine in other wordsMost
It is short, and load balancing degreesIt is minimum.
According to defining 9, for the Mission Scheduling that n different task is assigned to m different virtual machine is to find distribution
SchemeSo that the most long process time of virtual machineMost short and load balancing degreesIt is minimum.When only one of which is appointed
During the scheduling problem of business,.When there is the scheduling problem of k-1 task, 。
Theorem 1：For k task scheduling problem when, it is assumed that k-th task tkDistribute to z-th virtual machine vmz, i.e.,
Z virtual machine vmzTime span, and, then it is full
Sufficient recurrence relation。
Prove：Be given by definition 9Knowable to definition,
。
Can be obtained by theorem 1, k-th task tkThe virtual machine vm with earliest finish time will be distributed toz, i.e., complete earliest
The heuristic preferential allocation strategy of time.
The heuristic preferential allocation strategy based on earliest finish time of the present embodiment solves the stream of task scheduling initial solution
Journey figure is as shown in Fig. 2 step is as follows.
A01, the assignment instructions length and virtual machine execute instruction bar number per second be given based on Resources requirement model is calculated and appointed
Business setTInnIndividual task is in virtual machine setVM'smThe expected execution time on individual virtual machine, obtainn×mExpected execution
Time matrixC,Wherein it is expected execution time CijRepresent theiIndividual task isjThe time performed on individual virtual machine, equal to theiIt is individual
The command length of task is divided byjThe execute instruction bar number per second of individual virtual machine.
A02, initializes the present load array of m virtual machinevt[1..m] be zero, i.e., in the predecessor for starting distribution task
The present load of what virtual machine is zero.
A03, sequential access set of tasksTIn each task and execution step A04, A05 and A06, successively by thekIndividual
Business is distributed on the virtual machine with earliest finish time, until the distribution of all tasks obtains the initial of task scheduling after terminating
Solution.
A04, foundationmThe present load array of individual virtual machinevt[1..m] and expected execution time matrixC, calculatek
Individual taskt k Distribute time span corresponding to each virtual machinemakespan, wherein thejThe time span of individual virtual machine is thej
The present load array of individual virtual machinevt[j] and thekIndividual taskt k jThe execution time of individual virtual machinec kj Sum.
A05, finds out the minimum virtual machine of time spanvm x 。
A06, distributes taskt k To virtual machinevm x , updatevm x Virtual machine is loadedvt[x] for thexThe current of individual virtual machine is born
Carry arrayvt[x] and thekIndividual taskt k xThe execution time of individual virtual machinec kx Sum.
Define 10：For certain task allocative decision, it is assumed that vmxFor the virtual machine that load is maximum, vmyFor the void that load is minimum
Plan machine, i.e.,And；Assume k-th task tkIt is dispensed on virtual machine vmxOn hold
OK, l-th task tlIt is dispensed on virtual machine vmyUpper execution, i.e.,And.When task tkWith task tlHanded over
Change, i.e. task tkExchanged to virtual machine vmyUpper execution and task tlExchanged to virtual machine vmxUpper execution, virtual machine vmxExchange
The execution time difference in front and back is referred to as the financial value of the exchange。
The flow chart based on TABU search and greedy principle optimization task scheduling initial solution of the present embodiment as shown in figure 3,
Step is as follows.
B01, initialization task setTInnThe taboo array of individual tasktabu[1..n] be zero, that is, allow all of task
Exchanged.
B02, circulation execution step B03, B04, B05, B06, B07, B08, B09 and B10, the task pair for selecting financial value big
Swap, until the latest finishing time of general assignment cannot be reduced.
B03, is based onmThe load array of individual virtual machinevt[1..m], find out the maximum virtual machine of loadvm x With load most
Little virtual machinevm y 。
B04, if distributed to the maximum virtual machine of loadvm x All tasks be prohibited to exchange, then task switching terminates
And obtain the optimization solution of task scheduling；Otherwise in virtual machinevm x Select the allowing to be exchanged of the taskt k , by taskt k Correspondence
Taboo marktabu[kIt is revised as]=0tabu[k]=1。
B05, if distributed to the minimum virtual machine of loadvm y All of task is prohibited to exchange, then task switching terminates
And obtain the optimization solution of task scheduling；Otherwise in virtual machinevm y Calculate all the allowing to be exchanged of the tasks and taskt k Exchange it
Financial value afterwards.
B06, if the exchange financial value of all tasks pair for calculating is negative, skips step B07, B08, B09 and B10,
Continue cycling through execution step B01, otherwise execution step B07.
B07, select to exchange the maximum task of financial value to (t l , t k ) exchange, i.e. taskt k Exchanged to virtual machinevm y On
Perform and taskt l Exchanged to virtual machinevm x Upper execution.If virtual machine after exchangingvm y New load be more than virtual machinevm x 's
Original load, skips step B08, B09 and B10, continues cycling through execution step B01, otherwise execution step B08.
B08, changes taskt l Taboo marktabu[l]=1。
B09, updatesvm x Virtual machine is loadedvt[x]=vt[x]+c lx -c kx 。
B10, updatesvm y Virtual machine is loadedvt[y]=vt[y]+c ky -c ly 。
Claims (1)
1. it is a kind of based on TABU search and the cloud computing method for scheduling task of load balancing, it is characterised in that to comprise the following steps that：
Step 1, class types degree analysis is input into the task that user submits under cloud computing environment, and type and class degree are carried out to it
Analysis, determines the parallelization degree and feature of task；
Step 2, proceeding graininess decomposes, and is divided according to the parallelization degree and feature of user task, and the resource-sharing of cloud computing
With mode peculiar property, user task is decomposed according to proceeding graininess rank；
Step 3, resource characteristicses analysis, according to the resource-sharing method of salary distribution peculiar property of cloud computing, enters to the task after decomposition
Row resource characteristicses are analyzed；
Step 4, solves task scheduling initial solution, according to the analysis result to task resource characteristic, sets up and describes its resource requirement
Model, and then the initial solution of task scheduling is tried to achieve based on the model；
Step 5, optimizes task scheduling initial solution, optimizes task scheduling initial solution, shortens the latest finishing time of general assignment and changes
The load-balancing performance of kind virtual machine, obtains the optimization solution of task scheduling；
Step 6, duty mapping scheduling, by MapReduce Task Scheduling Models, the optimization solution of task scheduling is carried out mapping and
Scheduling, task of the realization in cloud computing environment is submitted to and performed, and effectively the load and shortening of balanced cloud computing platform is whole
The time span that task is completed；
It is as follows the step of described solution task scheduling initial solution in above-mentioned step 4；
Step 4.1, the assignment instructions length and virtual machine execute instruction bar number per second be given based on Resources requirement model is calculated and appointed
Business setTInnIndividual task is in virtual machine setVM'smThe expected execution time on individual virtual machine, obtainn×mExpected execution
Time matrixC,Wherein it is expected execution time CijRepresent theiIndividual task isjThe time performed on individual virtual machine, equal to theiIt is individual
The command length of task is divided byjThe execute instruction bar number per second of individual virtual machine；
Step 4.2, initializes the present load array of m virtual machinevt[1..m] be zero, i.e., in the predecessor for starting distribution task
The present load of what virtual machine is zero；
Step 4.3, sequential access set of tasksTIn each task, the heuristic preferential distribution plan based on earliest finish time
Slightly, successively by thekIndividual task is distributed on the virtual machine with earliest finish time, until the distribution of all tasks is obtained after terminating
The initial solution of task scheduling；
It is as follows the step of described optimization task scheduling initial solution in above-mentioned step 5；
Step 5.1, initialization task setTInnThe taboo array of individual tasktabu[1..n] be zero, that is, allow all of task
Exchanged；
Step 5.2, the big task of financial value is selected to swapping based on TABU search and greedy principle, until cannot reduce total
The latest finishing time of task；
In above-mentioned step 4.3, the described heuristic preferential allocation strategy based on earliest finish time is bykIndividual task distribution
It is as follows to step on the virtual machine with earliest finish time；
Step 4.3.1, foundationmThe present load array of individual virtual machinevt[1..m] and expected execution time matrixC, calculatekIndividual taskt k Distribute time span corresponding to each virtual machinemakespan, wherein thejThe time span of individual virtual machine is thejThe present load array of individual virtual machinevt[j] and thekIndividual taskt k jThe execution time of individual virtual machinec kj Sum；
Step 4.3.2, finds out the minimum virtual machine of time spanvm x ；
Step 4.3.3, distributes taskt k To virtual machinevm x , updatevm x Virtual machine is loadedvt[x] for thexIndividual virtual machine it is current
Load arrayvt[x] and thekIndividual taskt k xThe execution time of individual virtual machinec kx Sum；
In above-mentioned step 5.2, described selects the big task of financial value to swapping based on TABU search and greedy principle
Step is as follows；
Step 5.2.1, is based onmThe load array of individual virtual machinevt[1..m], find out the maximum virtual machine of loadvm x And load
Minimum virtual machinevm y ；
Step 5.2.2, if distributed to the maximum virtual machine of loadvm x All tasks be prohibited to exchange, then task switching terminates
And the optimization solution of task scheduling is obtained, go to step 6；Otherwise in virtual machinevm x Select the allowing to be exchanged of the taskt k , and will
The taskt k Corresponding taboo marktabu[kIt is revised as]=0tabu[k]=1；
Step 5.2.3, if distributed to the minimum virtual machine of loadvm y All of task is prohibited to exchange, then task switching terminates
And the optimization solution of task scheduling is obtained, go to step 6；Otherwise in virtual machinevm y Calculate all the allowing to be exchanged of the tasks and appoint
Businesst k Financial value after exchange；If the exchange financial value of all tasks pair for calculating is negative, skip step 5.2.4,
5.2.5,5.2.6 and 5.2.7, continues cycling through execution step 5.2.1, otherwise execution step 5.2.4；
Step 5.2.4, select to exchange the maximum task of financial value to (t l , t k ) exchange, i.e. taskt k Exchanged to virtual machinevm y
Upper execution and taskt l Exchanged to virtual machinevm x Upper execution；If virtual machine after exchangingvm y New load be more than virtual machinevm x
Former load, skip step 5.2.5,5.2.6 and 5.2.7, continue cycling through execution step 5.2.1, otherwise execution step 5.2.5；
Step 5.2.5, changes taskt l Taboo marktabu[l]=1；
Step 5.2.6, updatesvm x Virtual machine is loadedvt[x]=vt[x]+c lx -c kx ；
Step 5.2.7, updatesvm y Virtual machine is loadedvt[y]=vt[y]+c ky -c ly 。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410527189.XA CN104331321B (en) | 2014-10-09 | 2014-10-09 | Cloud computing task scheduling method based on tabu search and load balancing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410527189.XA CN104331321B (en) | 2014-10-09 | 2014-10-09 | Cloud computing task scheduling method based on tabu search and load balancing |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104331321A CN104331321A (en) | 2015-02-04 |
| CN104331321B true CN104331321B (en) | 2017-04-19 |
Family
ID=52406055
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410527189.XA Expired - Fee Related CN104331321B (en) | 2014-10-09 | 2014-10-09 | Cloud computing task scheduling method based on tabu search and load balancing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104331321B (en) |
Families Citing this family (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105159762B (en) * | 2015-08-03 | 2018-09-07 | 冷子阳 | Heuristic cloud computing method for scheduling task based on Greedy strategy |
| CN105468453A (en) * | 2015-08-03 | 2016-04-06 | 汉柏科技有限公司 | Vector scheduling method and apparatus for cloud computing system |
| US10055262B1 (en) | 2015-12-11 | 2018-08-21 | Amazon Technologies, Inc. | Distributed load balancing with imperfect workload information |
| CN107617216B (en) * | 2016-07-15 | 2020-10-16 | 珠海金山网络游戏科技有限公司 | System and method for designing game artificial intelligence task |
| CN106302224A (en) * | 2016-10-15 | 2017-01-04 | 黄林果 | A kind of task optimization bootstrap technique under cloud service environment |
| CN106991006B (en) * | 2017-03-30 | 2018-05-15 | 浙江天正信息科技有限公司 | Support the cloud workflow task clustering method relied on and the time balances |
| CN107329815B (en) * | 2017-06-26 | 2020-08-04 | 浙江工业大学 | BP-Tabu search-based cloud task load balancing scheduling method |
| CN109785931A (en) * | 2017-11-10 | 2019-05-21 | 北京连心医疗科技有限公司 | Based on Optimum distribution formula cloud radiotherapy planning system and application method, storage medium |
| CN108108228A (en) * | 2018-01-05 | 2018-06-01 | 安徽师范大学 | A kind of resource allocation methods based on differential evolution algorithm |
| CN108170861B (en) * | 2018-01-25 | 2020-07-03 | 中润普达(十堰)大数据中心有限公司 | Distributed database system collaborative optimization method based on dynamic programming |
| CN110362388B (en) * | 2018-04-11 | 2021-08-31 | 中移(苏州)软件技术有限公司 | Resource scheduling method and device |
| CN109194746B (en) * | 2018-09-06 | 2021-03-26 | 贵州我联你联网络技术有限公司 | Heterogeneous information processing method based on Internet of things |
| CN109948944B (en) * | 2019-03-27 | 2021-04-06 | 中南林业科技大学 | Satellite task scheduling method and system |
| CN110311965A (en) * | 2019-06-21 | 2019-10-08 | 长沙学院 | Method for scheduling task and system under a kind of cloud computing environment |
| CN110489229B (en) * | 2019-07-17 | 2020-06-30 | 长沙学院 | Multi-target task scheduling method and system |
| CN110688204B (en) * | 2019-08-08 | 2022-08-26 | 平安科技（深圳）有限公司 | Distributed computing system task allocation method and related equipment |
| CN112181658B (en) * | 2020-09-30 | 2024-04-05 | 南京工程学院 | Calculation task allocation method for maximizing network benefits in heterogeneous network |
| CN112764879A (en) * | 2021-01-14 | 2021-05-07 | 深圳市科思科技股份有限公司 | Load balancing method, electronic device and computer readable storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663216A (en) * | 2012-05-16 | 2012-09-12 | 孙凌宇 | Core value calculating method of large-scale integrated circuit based on node attribute function |
| CN102682176A (en) * | 2012-05-18 | 2012-09-19 | 冷明 | Method for dividing large-scale integrated circuit based on cellular automaton and empowerment hypergraph |
-
2014
- 2014-10-09 CN CN201410527189.XA patent/CN104331321B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663216A (en) * | 2012-05-16 | 2012-09-12 | 孙凌宇 | Core value calculating method of large-scale integrated circuit based on node attribute function |
| CN102682176A (en) * | 2012-05-18 | 2012-09-19 | 冷明 | Method for dividing large-scale integrated circuit based on cellular automaton and empowerment hypergraph |
Non-Patent Citations (2)
| Title |
|---|
| 一种基于元胞自动机的无向图剖分优化算法;孙凌宇、冷明等;《计算机工程与应用》;20081231;第44卷(第24期);全文 * |
| 赋权有向图的最小生成树算法;孙凌宇、冷明等;《计算机工程与应用》;20100131;第36卷(第2期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104331321A (en) | 2015-02-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104331321B (en) | Cloud computing task scheduling method based on tabu search and load balancing | |
| CN105159762B (en) | Heuristic cloud computing method for scheduling task based on Greedy strategy | |
| Kaur et al. | Container-as-a-service at the edge: Trade-off between energy efficiency and service availability at fog nano data centers | |
| Lin et al. | Fast GA-based project scheduling for computing resources allocation in a cloud manufacturing system | |
| Liu et al. | Job scheduling model for cloud computing based on multi-objective genetic algorithm | |
| CN103729246B (en) | Method and device for dispatching tasks | |
| US8527988B1 (en) | Proximity mapping of virtual-machine threads to processors | |
| CN100576177C (en) | Bidirectional grade gridding resource scheduling method based on the QoS constraint | |
| CN106371924B (en) | A kind of method for scheduling task minimizing MapReduce cluster energy consumption | |
| CN107329815A (en) | A kind of cloud task load equalization scheduling method searched for based on BP Tabu | |
| CN105005503B (en) | Cloud computing load balancing method for scheduling task based on cellular automata | |
| CN102968344A (en) | Method for migration scheduling of multiple virtual machines | |
| CN105373432B (en) | A kind of cloud computing resource scheduling method based on virtual resource status predication | |
| CN104050042A (en) | Resource allocation method and resource allocation device for ETL (Extraction-Transformation-Loading) jobs | |
| Tantalaki et al. | Pipeline-based linear scheduling of big data streams in the cloud | |
| Li et al. | An effective scheduling strategy based on hypergraph partition in geographically distributed datacenters | |
| CN108427602B (en) | Distributed computing task cooperative scheduling method and device | |
| CN111309472A (en) | Online virtual resource allocation method based on virtual machine pre-deployment | |
| CN104917839A (en) | Load balancing method for use in cloud computing environment | |
| CN104112049A (en) | P2P (peer-to-peer) architecture based cross-data-center MapReduce task scheduling system and P2P architecture based cross-data-center MapReduce task scheduling method | |
| CN110187970A (en) | A kind of distributed big data parallel calculating method based on Hadoop MapReduce | |
| Geng et al. | A task scheduling algorithm based on priority list and task duplication in cloud computing environment | |
| Li et al. | A wholistic optimization of containerized workflow scheduling and deployment in the cloud–edge environment | |
| CN109710372A (en) | A kind of computation-intensive cloud workflow schedule method based on cat owl searching algorithm | |
| Rodrigues et al. | Network-aware container scheduling in multi-tenant data center |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170419 Termination date: 20171009 |