CN109710372B - Calculation intensive cloud workflow scheduling method based on owl search algorithm - Google Patents
Calculation intensive cloud workflow scheduling method based on owl search algorithm Download PDFInfo
- Publication number
- CN109710372B CN109710372B CN201811336040.8A CN201811336040A CN109710372B CN 109710372 B CN109710372 B CN 109710372B CN 201811336040 A CN201811336040 A CN 201811336040A CN 109710372 B CN109710372 B CN 109710372B
- Authority
- CN
- China
- Prior art keywords
- task
- scheduling
- time
- cloud workflow
- workflow
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000010845 search algorithm Methods 0.000 title claims abstract description 24
- 238000004364 calculation method Methods 0.000 title description 3
- 238000005457 optimization Methods 0.000 claims abstract description 17
- 230000007614 genetic variation Effects 0.000 claims abstract description 5
- 238000004422 calculation algorithm Methods 0.000 claims description 38
- 238000013507 mapping Methods 0.000 claims description 8
- 230000005540 biological transmission Effects 0.000 claims description 4
- 238000004891 communication Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 230000001419 dependent effect Effects 0.000 claims description 2
- 230000007547 defect Effects 0.000 abstract description 2
- 230000002068 genetic effect Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000012804 iterative process Methods 0.000 description 2
- 239000002245 particle Substances 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000010076 replication Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a computing-intensive cloud workflow scheduling method based on a owl search algorithm, and belongs to the technical field of cloud computing. Modifying a population iterative update formula in a owl search algorithm to update each scheduling scheme according to the influence of the optimal scheduling scheme on the optimal scheduling scheme, so that optimization is more targeted; in a population iteration updating mechanism, randomness is introduced by utilizing a genetic variation thought, so that the search process is prevented from falling into local optimum, an optimum scheduling scheme can be obtained in a shorter time, the virtual machines are reasonably distributed, and tasks are efficiently scheduled. The method can effectively overcome the defects of high randomness, easy trapping in local optimization and low convergence speed of optimal solution search in the prior art, improves the search efficiency, shortens the search time, can find a more optimal scheduling scheme in a shorter time, and reduces the time overhead of workflow scheduling.
Description
Technical Field
The invention relates to a cloud workflow scheduling method, in particular to a computing-intensive cloud workflow scheduling method based on a owl search algorithm, and belongs to the technical field of cloud computing.
Background
Cloud computing is used as a business computing mode, a virtualization technology is adopted, resources such as storage, computing and network communication of a data center are integrated into a shared computing resource pool which can be dynamically configured, and computing services paid by use are provided for users. Users can access the configurable shared computing resource pool (such as servers, storage, application software, networks and the like) through available and convenient networks without purchasing any hardware resources such as servers and the like, and obtain computing capacity, storage space and information services as required.
With the continuous development of cloud computing, large-scale complex workflows become a new mode of cloud computing application. The execution of the cloud workflow mainly comprises two stages of task scheduling and resource supply. In the task scheduling process, a proper virtual machine needs to be selected for the task requested by the user according to a proper scheduling strategy, and the constraints of quality of service (QoS) and the like are met, so that the whole scheduling process is completed. The computing-intensive cloud workflow is composed of a plurality of subtasks with mutual dependency relationships, so that in the whole workflow scheduling process, not only the execution time of tasks needs to be considered, but also the dependency relationship constraint among the tasks needs to be satisfied, and the execution span time (makespan) of the whole workflow is shortest. Different task allocation strategies can directly influence the execution time and cost of the whole cloud workflow, how to allocate the most appropriate computing resources for the cloud workflow tasks, and achieve the scheduling objective while meeting the logic dependence constraint among the tasks, and become a problem to be solved urgently by various cloud service providers.
Cloud workflow scheduling is a typical NP-hard problem, and at present, a heuristic algorithm and a random search algorithm are mainly adopted for solving. Heuristic algorithms are mainly divided into list scheduling, task replication, task set clustering and the like, such as HEFT, MIN-MIN and MIN-MAX, and near-optimal solutions are not easy to find. The random search algorithm mainly comprises a genetic algorithm, a particle swarm algorithm, an immune evolution algorithm and the like, and is essentially designed to be an efficient search strategy. The evolutionary algorithms such as the genetic algorithm have the advantages of global search and the capability of avoiding falling into local optimum, but the search time is too long, so that the real-time performance of the algorithm is influenced; the swarm intelligence optimization algorithm has the advantages of high convergence speed and wide application range, but lacks an effective local search mechanism.
Disclosure of Invention
The invention aims to solve the problem of computing-intensive cloud work scheduling and provides a computing-intensive cloud workflow scheduling method based on an improved owl search algorithm. The basic idea is as follows: and traversing and searching different scheduling schemes depending on mapping from tasks to virtual machine resources in the cloud workflow by adopting a owl search algorithm, and searching the scheduling scheme with the minimum workflow execution span time. Meanwhile, according to the characteristics of the cloud workflow, the existing owl search algorithm is improved, firstly, the intensity variation is defined as the influence of the optimal solution (namely the optimal scheduling scheme) on other different individuals (or the scheduling scheme) through the inverse sound intensity square law, and the optimization step length of the different individuals is adaptively adjusted, so that the search efficiency of the optimal solution is greatly improved; secondly, aiming at the characteristics of the cloud workflow scheduling problem, the optimization direction of the individual is modified to avoid generating excessive invalid solutions, and all the individuals are enabled to directly and gradually approach to the optimal solution according to different step lengths, so that the stability of the individual solution is improved, and the optimization speed of the whole algorithm is improved; and thirdly, aiming at the problem that the group intelligent optimization algorithm is easy to fall into local optimum, introducing randomness by adding a variation strategy in a group iteration updating mechanism by utilizing the variation thought of evolutionary computation, and randomly changing the mapping relation between individual tasks and virtual machines in some scheduling schemes when the optimal solution is not updated for one iteration, so as to jump out the local optimum and search a better overall scheduling scheme.
The method comprises the following steps:
step one, inputting a to-be-scheduled computing intensive cloud workflow model submitted by a user and a dependent subtask set contained in the to-be-scheduled computing intensive cloud workflow model, and a virtual machine set available for leasing;
and step two, scheduling each cloud workflow subtask to the most suitable virtual machine to execute, and modeling as a standard minimum value solving problem. The scheduling targets are as follows: and optimizing the execution span time makespan of the whole cloud workflow to minimize the time spent by all cloud workflow tasks after the execution.
And step three, solving the task-virtual machine scheduling problem in the cloud computing environment by using a owl search algorithm based on the inverse square law of sound intensity. The iterative process comprises the following steps:
step 2, initializing each scheduling scheme by using uniformly distributed random numbers;
and 5, finding the optimal scheduling scheme in the current generation. bestNum ═ t if the optimal solution has updates. Updating the distance information between each scheduling scheme and the current optimal solution and the intensity variable quantity of each scheduling scheme;
and 8, finding an optimal scheduling scheme, and binding the workflow subtasks and the virtual machines according to the mapping relation between the tasks and the virtual machines given by the scheduling scheme.
Advantageous effects
The method can effectively overcome the defects of high randomness, easy trapping in local optimization and low convergence speed of optimal solution search in the prior art, improves the search efficiency, shortens the search time, can find a more optimal scheduling scheme in a shorter time, reduces the total time overhead of workflow scheduling, and specifically comprises the following three points:
1. the owl search algorithm is applied to the scheduling problem for the first time, and a new solution is provided for the computation-intensive cloud workflow scheduling.
2. By improving the existing owl search algorithm, namely modifying the population update iterative formula, the randomness of the optimal solution search is effectively reduced, the search efficiency is improved, and the optimization process has better target guidance.
3. By applying the genetic variation thought and introducing randomness in the population iteration updating mechanism of the existing owl search algorithm, the situation that the search is trapped in local optimum is effectively avoided, a global optimal scheduling scheme can be found in a shorter time, the convergence speed of the algorithm is improved, and the scheduling performance of the cloud workflow is improved.
Drawings
Fig. 1 is a flow of a cloud workflow scheduling method based on a owl search algorithm according to the present invention.
FIG. 2 is a simple 17 task Montage workflow.
FIG. 3 is a process for executing span time minimum value change for Montage _25 by different algorithms.
FIG. 4 is a process of executing span time minimum value change for Montage _50 by different algorithms.
FIG. 5 shows the span time minimum variation process performed by different algorithms for Montage _ 100.
FIG. 6 is a process of optimizing for Montage _25 by different algorithms.
FIG. 7 is a process of optimizing for Montage _50 by different algorithms.
FIG. 8 is a process of optimizing for Montage _100 by different algorithms.
Fig. 9 shows the program run times for the different algorithms.
FIG. 10 is a minimum value of the execution span time of the optimal scheduling schemes of different algorithms.
Detailed Description
The method of the present invention will be described in detail below by way of examples with reference to the accompanying drawings.
A computing-intensive cloud workflow scheduling method based on a owl search algorithm is shown in FIG. 1 and comprises the following steps:
inputting a to-be-scheduled calculation intensive workflow model submitted by a user and a corresponding dependency subtask set thereof, and a virtual machine resource set available for leasing;
for a compute intensive cloud workflow scheduling problem, the cloud workflow is described as a directed acyclic graph G ═ T, E, where: t is a set of nodes in the directed acyclic graph, representing n tasks in the cloud workflow, i.e., T ═ T { (T)1,T2,……,Ti,Tj,……,TnWhere i, j ═ 1,2, … …, n; t isentryFor entry tasks, TexitIs an export task; e is the set of directed edges in the directed acyclic graph, E { < T {i,Tj＞|Ti,Tj∈ T, and has a directed edge Ti→TjRepresenting a parent task TiAnd its subtask TjDependency between them, TjOnly in its parent task TiExecution can only begin after completion. As shown in fig. 2, a simple Montage workflow with 17 tasks, that is, the number of tasks n is 17, directed arcs (arrowed lines) in the cloud workflow graph represent dependencies E between tasks, and corresponding numbers on the directed arcs represent sizes of files to be transmitted between parent and child tasks.
Using VM to represent virtual machines, m represents the total number of virtual machines which can be rented by users, and the virtual machine resource set can be represented as: VM ═ VM1,VM2,……,VMk,……,VMmWhere k is 1, 2. Assuming the MIPS represents the number of million machine language instructions per second that a computing device can process, the virtual machine VMkAvailable MIPS (VM)k) To indicate.
And step two, scheduling each cloud workflow subtask to the most suitable virtual machine to execute, and modeling as a standard minimum value solving problem. The method comprises the following specific steps:
suppose that: (1) all subtasks in the task set are atomic tasks, namely, each task can not be split into tasks with smaller granularity; (2) each virtual machine can only process one task at the same time, namely, only when the virtual machine finishes the task currently processed, the request of a new task can be received; (3) task execution is not interruptible, i.e., each subtask is not allowed to be interrupted by other task requests while executing or performing computations on the virtual machine it is rented.
The scheduling objective is to optimize the execution span time overhead of the whole cloud workflow task, namely to make the total time makespan spent on all cloud workflow subtasks being executed shortest;
the constraint condition is that the number n of the cloud workflow subtasks is larger than the number m of the virtual machines available for leasing, namely n is larger than m;
defining a task TiIn a virtual machine VMkExecution time on ETC (T)i,VMk) And a parent task TiAnd subtask TjTransmission time TT (T) betweeni,Tj) The following were used:
among them, Length (T)i) Representing a task TiInstruction length of (3), MIPS (VM)k) Representing virtual machines VMkThe processing speed of (2); TransferSize (T)i,Tj) Representing a parent task TiAnd subtask TjThe bandwidth of the communication line between the virtual machines is represented by the bandwidth of the transmission file between the virtual machines.
Defining task T in cloud workflowiStart time ST (T)i) And a completion time FT (T)i) The following were used:
FT(Ti)＝ST(Ti)+ETC(Ti,VMk) (4)
wherein, ST (T)entry) Representing an ingress task TentryStart time of (D), ST (T)i) Representing a task TiStart time of (2), FT (T)p)、FT(Ti) Respectively represent tasks TpAnd its subtask TiCompletion time of (d), avail (VM)k) Representing virtual machines VMkAvailable time of (D), predr (T)i) Representing a task TiIs a set of all parent tasks, TT (T)p,Ti) Representing a task TpAnd its subtask TiThe transmission time in between.
The total execution span time overhead makespan of the cloud workflow is represented by the maximum value of the completion time of all subtasks in the cloud workflow, that is:
thirdly, solving a cloud workflow task-virtual machine scheduling problem by using a owl search algorithm based on the inverse square law of sound intensity, wherein the iterative process comprises the following steps:
step 2, generating M initial scheduling schemes of 0 th generation by using uniformly distributed random numbers
Wherein s 1,2, M:
wherein, OLAnd OURespectively, the 0 th generation s scheduling scheme
Middle task TiThe lower and upper limits of the numbers, U (0,1) being the interval [0,1 ]]Random numbers uniformly distributed within the range.
wherein,
for task TiIn a scheduling scheme
The following completion time.
And 5, finding the optimal scheduling scheme in the current generation. bestNum ═ t if the optimal solution has updates. For each scheduling scheme
Updating
Distance information from the current optimal solution
Simultaneous update
Amount of intensity variation of
In the formula, V represents a global optimal scheduling scheme, makespan (V) represents workflow execution span time corresponding to the optimal scheduling scheme, and random represents a random number of [0,1 ];
and 6, judging whether the generation of the iteration is I and the optimal solution is not updated. If t-bestNum > l, randomly changing the task-virtual machine mapping relation of any position of any individual by using the thought of genetic variation, and then turning to the step 7; otherwise, directly turning to the step 7;
wherein,
represents the s-th scheduling scheme of the t-th generation.
And 8, finding an optimal scheduling scheme, and binding the workflow subtasks and the virtual machines according to the mapping relation between the tasks and the virtual machines given by the scheduling scheme.
Examples
In order to test the effect of utilizing the improved Owl Search Algorithm (OSA) to schedule the cloud workflow, the cloud computing simulation tool WorkflowSim is used for simulating a cloud computing data center, and the computing efficiency of the makespan is improved by optimizing the estimation algorithm of the workflow execution span time makespan. The most common intelligent optimization algorithms such as Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) are selected for comparison in the experiment.
For different scale Montage workflow models, 10 same virtual machines are used for carrying out experiments respectively, the minimum execution span time of the workflow and the average execution span time of the workflow corresponding to each generation of individuals are selected as scheduling performance indexes to measure the generalization capability and performance of the algorithm, and comparison results are shown in fig. 3 to fig. 10.
As can be seen from fig. 3, 4, and 5, for the workflow models of different scales, the scheduling algorithm based on the search of the owl can find a better solution, and an approximately optimal solution can be found with fewer iterations. As can be seen from fig. 6, 7, and 8, the OSA algorithm has a fast optimizing speed and is not prone to fall into local optimization. As can be seen from fig. 9, the algorithm of the OSA is less time-complex. As can be seen from FIG. 10, for small workflows, the OSA algorithm can find a better solution; for medium and large workflows, the improved OSA algorithm of the present invention can find a better scheduling scheme with execution spans of time 95.03, 174.15, respectively, which is the same as the optimization result of the GA algorithm. However, as can be seen from fig. 9, the execution time overhead of the GA algorithm is large, which is more than 2 times that of the OSA algorithm.
Claims (5)
1. A computing intensive cloud workflow scheduling method based on a owl search algorithm is characterized by comprising the following steps:
step one, inputting a to-be-scheduled computing intensive cloud workflow model submitted by a user, a dependent subtask set contained in the to-be-scheduled computing intensive cloud workflow model and a virtual machine set available for lease;
step two, scheduling each cloud workflow subtask to the process executed on the most suitable virtual machine, modeling as a standard minimum solving problem, and scheduling the tasks with the scheduling targets as follows: optimizing the execution span time makespan of the whole cloud workflow to minimize the time spent by all cloud workflow tasks after the execution;
thirdly, solving a task-virtual machine scheduling problem in the cloud computing environment by using a owl search algorithm based on the inverse square law of sound intensity, comprising the following steps of:
step 1, initializing basic parameters of an algorithm, wherein the basic parameters comprise a step length parameter beta, the number M of scheduling schemes, the maximum Iteration number Iteraction and the optimization Iteration number bestNum when an optimal solution appears for the first time;
step 2, initializing each scheduling scheme by using uniformly distributed random numbers;
step 3, in the algorithm Iteration process, when the Iteration time t is less than the maximum Iteration time Iteration, t is t +1, and step 4 is switched; when the Iteration time t is greater than or equal to the maximum Iteration time Iteration, turning to step 8;
step 4, calculating workflow execution span time makespan of all scheduling schemes in the current generation according to a cloud workflow model, namely the dependency relationship among subtasks;
step 5, finding the optimal scheduling scheme in the current generation; if the optimal solution is updated, bestNum is t; for each scheduling scheme
Updating
Distance information from the current optimal solution
Simultaneous update
Amount of intensity variation of
Wherein, V represents a global optimal scheduling scheme, makespan (V) represents a workflow execution span time corresponding to the optimal scheduling scheme, and random represents a random number of [0,1 ];
step 6, judging whether l generations of iteration is performed and the optimal solution is not updated, if t-bestNum > l, randomly changing the mapping relation of the virtual machine at any position of any individual by using the idea of genetic variation, and turning to step 7; otherwise, directly turning to the step 7;
step 7, updating all scheduling schemes of the current generation according to the intensity variation, and returning to the step 3;
and 8, finding an optimal scheduling scheme, and binding the workflow subtasks and the virtual machines according to the mapping relation between the tasks and the virtual machines given by the scheduling scheme.
2. The method for scheduling a computing-intensive cloud workflow based on the owl search algorithm according to claim 1, wherein the first step describes the cloud workflow specifically as a directed acyclic graph G ═ (T, E), where: t is a set of nodes in the directed acyclic graph and represents n tasks in the cloud workflow, and T is { T ═ T { (T)1,T2,……,Tn}; e is the set of directed edges in the cloud workflow model, E { (T)i,Tj)|Ti,Tj∈ T, and has a directed edge Ti→TjRepresenting a parent task TiAnd subtask TjDependency between them, TjOnly at TiCan be executed after completion; the virtual machines are represented by VM, m represents the total number of the virtual machines which can be rented by a user, and the virtual machine resource set is represented as VM ═ VM1,VM2,……,VMk,……,VMmWhere k is 1, 2.., m; MIPS represents the number of million machine language instructions per second that a computing device can process, then MIPS (VM)k) Representing virtual machines VMkThe processing speed of (2).
3. The computing-intensive cloud workflow scheduling method based on the owl search algorithm according to claim 1, wherein the modeling method of the second step is:
all subtasks in the task set are designed to be atomic tasks, namely, each task can not be split into tasks with smaller granularity; each virtual machine can only process one task at the same time, namely, only when the virtual machine finishes the task currently processed, the request of a new task can be received; the task execution can not be interrupted, namely, each subtask is not allowed to be interrupted by other task requests when being executed or calculated on the virtual machine rented by the subtask;
the constraint condition is that the number n of the cloud workflow subtasks is larger than the number m of the virtual machines available for leasing, namely n is larger than m;
defining a task TiIn a virtual machine VMkExecution time on ETC (T)i,VMk) And a parent task TiAnd subtask TjTransmission time TT (T) betweeni,Tj) The following were used:
among them, Length (T)i) Representing a task TiInstruction length of (3), MIPS (VM)k) Representing virtual machines VMkThe processing speed of (2); TransferSize (T)i,Tj) Representing a parent task TiAnd subtask TjThe bandwidth of a communication line between the virtual machines is represented by the bandwidth of the transmission file between the virtual machines; the cloud workflow is described as a directed acyclic graph G ═ (T, E), where T is a set of nodes in the directed acyclic graph, representing n tasks in the cloud workflow, and T ═ T { (T, E)1,T2,……,Tn}；
Defining task T in cloud workflowiStart time ST (T)i) And a completion time FT (T)i) The following were used:
FT(Ti)＝ST(Ti)+ETC(Ti,VMk) (4)
wherein, ST (T)entry) Representing an ingress task TentryStart time of (D), ST (T)i) Representing a task TiStart time of (2), FT (T)p)、FT(Ti) Respectively represent tasks TpAnd its subtask TiCompletion time of (d), avail (VM)k) Representing virtual machines VMkAvailable time of (D), predr (T)i) Representing a task TiIs a set of all parent tasks, TT (T)p,Ti) Representing a task TpAnd its subtask TiThe transmission time therebetween;
the execution span time overhead of the cloud workflow is represented by the maximum value of the completion time of all the subtasks in the cloud workflow, namely:
wherein makespan represents the execution span time overhead of the cloud workflow.
4. The computing-intensive cloud workflow scheduling method based on the owl search algorithm according to claim 1, wherein the method for generating each initial scheduling scheme in the step 2 in the step three is as follows: generation of M initial scheduling schemes of generation 0 using uniformly distributed random numbers
Wherein s 1,2, M:
wherein, OLAnd OURespectively, the 0 th generation s scheduling scheme
Middle task TiThe lower and upper limits of the numbers, U (0,1) being the interval [0,1 ]]Random numbers uniformly distributed within the range.
5. The computing-intensive cloud workflow scheduling method based on the owl search algorithm according to claim 1, wherein the updating method of step 7 in the third step is:
wherein,
denotes the ith scheduling scheme of the t-th generation, β denotes a step size parameter,
indicating the intensity variation of each scheduling scheme, and V indicating a global optimal scheduling scheme.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811336040.8A CN109710372B (en) | 2018-11-12 | 2018-11-12 | Calculation intensive cloud workflow scheduling method based on owl search algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811336040.8A CN109710372B (en) | 2018-11-12 | 2018-11-12 | Calculation intensive cloud workflow scheduling method based on owl search algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109710372A CN109710372A (en) | 2019-05-03 |
| CN109710372B true CN109710372B (en) | 2020-10-13 |
Family
ID=66254212
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811336040.8A Active CN109710372B (en) | 2018-11-12 | 2018-11-12 | Calculation intensive cloud workflow scheduling method based on owl search algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109710372B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113157381B (en) * | 2020-05-14 | 2022-07-05 | 北京理工大学 | Multi-workflow scheduling method based on non-dominated sorting and owl search |
| CN112685138B (en) * | 2021-01-08 | 2022-08-23 | 北京理工大学 | Multi-workflow scheduling method based on multi-population hybrid intelligent optimization in cloud environment |
| CN113220414B (en) * | 2021-04-22 | 2022-09-09 | 北京理工大学 | Cloud workflow scheduling method based on improved lean-rich optimization algorithm |
| CN112905307B (en) * | 2021-04-22 | 2022-07-22 | 北京理工大学 | Multi-target cloud workflow scheduling method based on lean and rich optimization algorithm |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103067524A (en) * | 2013-01-18 | 2013-04-24 | 浪潮电子信息产业股份有限公司 | Ant colony optimization computing resource distribution method based on cloud computing environment |
| CN108133260A (en) * | 2018-01-17 | 2018-06-08 | 浙江理工大学 | The workflow schedule method of multi-objective particle swarm optimization based on real-time status monitoring |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8296765B2 (en) * | 2010-07-27 | 2012-10-23 | Kurdi Heba A | Method of forming a personal mobile grid system and resource scheduling thereon |
| CN103412792B (en) * | 2013-07-18 | 2015-06-10 | 成都国科海博信息技术股份有限公司 | Dynamic task scheduling method and device under cloud computing platform environment |
| CN106648831B (en) * | 2016-12-08 | 2019-11-05 | 杭州电子科技大学 | Cloud workflow schedule method based on glowworm swarm algorithm and dynamic priority |
-
2018
- 2018-11-12 CN CN201811336040.8A patent/CN109710372B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103067524A (en) * | 2013-01-18 | 2013-04-24 | 浪潮电子信息产业股份有限公司 | Ant colony optimization computing resource distribution method based on cloud computing environment |
| CN108133260A (en) * | 2018-01-17 | 2018-06-08 | 浙江理工大学 | The workflow schedule method of multi-objective particle swarm optimization based on real-time status monitoring |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109710372A (en) | 2019-05-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Tong et al. | QL-HEFT: a novel machine learning scheduling scheme base on cloud computing environment | |
| Cheng et al. | Cost-aware job scheduling for cloud instances using deep reinforcement learning | |
| Jayanetti et al. | Deep reinforcement learning for energy and time optimized scheduling of precedence-constrained tasks in edge–cloud computing environments | |
| Zuo et al. | A multi-objective hybrid cloud resource scheduling method based on deadline and cost constraints | |
| CN109710372B (en) | Calculation intensive cloud workflow scheduling method based on owl search algorithm | |
| Kamalinia et al. | Hybrid task scheduling method for cloud computing by genetic and DE algorithms | |
| CN108804227A (en) | The method of the unloading of computation-intensive task and best resource configuration based on mobile cloud computing | |
| Stavrinides et al. | Orchestrating real-time IoT workflows in a fog computing environment utilizing partial computations with end-to-end error propagation | |
| Xiao et al. | Learning non-cooperative game for load balancing under self-interested distributed environment | |
| Tos et al. | A performance and profit oriented data replication strategy for cloud systems | |
| Tong et al. | DDQN-TS: A novel bi-objective intelligent scheduling algorithm in the cloud environment | |
| Hosseini | A new shuffled genetic-based task scheduling algorithm in heterogeneous distributed systems | |
| Cheng et al. | Cost-aware real-time job scheduling for hybrid cloud using deep reinforcement learning | |
| Elsedimy et al. | MOTS‐ACO: An improved ant colony optimiser for multi‐objective task scheduling optimisation problem in cloud data centres | |
| Rafsanjani et al. | A new heuristic approach for scheduling independent tasks on heterogeneous computing systems | |
| Kousalya et al. | To improve ant algorithm’s grid scheduling using local search | |
| Nazeri et al. | A predictive energy-aware scheduling strategy for scientific workflows in fog computing | |
| Kumar et al. | EAEFA: An Efficient Energy-Aware Task Scheduling in Cloud Environment | |
| Huang et al. | Computation offloading for multimedia workflows with deadline constraints in cloudlet-based mobile cloud | |
| Rawat et al. | Learning-based task scheduling using big bang big crunch for cloud computing environment | |
| Komarasamy et al. | ScHeduling of jobs and Adaptive Resource Provisioning (SHARP) approach in cloud computing | |
| Zhang et al. | Data-intensive workflow scheduling strategy based on deep reinforcement learning in multi-clouds | |
| CN117687759A (en) | Task scheduling method, device, processing equipment and readable storage medium | |
| Shubair | Enhancement of task scheduling technique of big data cloud computing | |
| Lin et al. | Joint deadline-constrained and influence-aware design for allocating MapReduce jobs in cloud computing systems |
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 |