CN110264097A - More scientific workflows based on cloud environment concurrently execute dispatching method - Google Patents
More scientific workflows based on cloud environment concurrently execute dispatching method Download PDFInfo
- Publication number
- CN110264097A CN110264097A CN201910557811.4A CN201910557811A CN110264097A CN 110264097 A CN110264097 A CN 110264097A CN 201910557811 A CN201910557811 A CN 201910557811A CN 110264097 A CN110264097 A CN 110264097A
- Authority
- CN
- China
- Prior art keywords
- task
- server
- list
- time
- cloud environment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0633—Workflow analysis
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- Educational Administration (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
More scientific workflows based on cloud environment concurrently execute dispatching method.So-called Multi-workflow, which concurrently executes, refers to that multiple workflows are performed simultaneously, and the task in each workflow is interspersed to be executed.List workflow execution scheduling research at present is more, has ignored the waiting time of server, not only causes the waste of resource, and increases the waiting time of follow-up work stream.The method of the present invention includes the following steps: that reduction scheduling length achievees the effect that shorten task completion time firstly, passing through the calculative strategy of task order.Then, in order to guarantee to dispatch the fairness in implementation procedure, the strategy of building task pool is proposed；Secondly, the task in task pool is according to priority passed through resource selection strategy, reduce task completion time, improves the utilization rate of resource；Finally, reaching reduces the target for executing energy consumption to there are the resources of free time section to carry out voltage adjustment in such a way that dynamic voltage frequency adjusts.The present invention concurrently executes scheduling for more scientific workflows based on cloud environment.
Description
Technical field
The present invention relates to a kind of, and more scientific workflows based on cloud environment concurrently execute dispatching method.
Background technique
Multi-workflow, which concurrently executes, refers to that multiple workflows are performed simultaneously, and the task in each workflow is interspersed to be executed.At present
Most popular dispatching method is critical path (CPOP) method in isomery earliest finish time (HEFT) method and processor；
HEFT method is operated based on the earliest time that available resources execute, and has ignored the other factors for influencing task execution time,
Two inter-related tasks may be distributed to different resources, increase the execution length of task；CPOP method be using task to
Upper and downward ranking summation is shortened by hierarchical selection resource to calculate the grade of each task and is executed the time.But CPOP
The implementation schedule of method is now not so good as HEFT method.
Summary of the invention
The purpose of the present invention is to solve more scientific workflows under current cloud environment concurrently to execute scheduling problem, to reduce
Task completion time improves resource utilization, and reducing and executing energy consumption is target, provides a kind of more research-on-researches based on cloud environment
Stream concurrently executes dispatching method.
Above-mentioned purpose is realized by following technical scheme:
A kind of more scientific workflows based on cloud environment concurrently execute dispatching method, it is characterized in that: using optimization mould is concurrently executed
Type.1. using the calculative strategy of task order, reduce scheduling length, achievees the effect that shorten task completion time；2. using structure
The strategy for building task pool ensure that the fairness in scheduling implementation procedure；3. according to priority being passed through using the task in task pool
Resource selection strategy reduces task completion time, improves the utilization rate of resource；4. the side adjusted by dynamic voltage frequency
Formula, reaching reduces the target for executing energy consumption.
More scientific workflows of the consideration based on cloud environment according to claim 1 concurrently execute dispatching method, described
Dispatching method specific implementation step it is as follows:
Step 1: each task node information of each scientific workflow being stored in one and is initialized as in empty queue GQueue；
Step 2: initialization ListRFor sky, it is used to store the collating sequence of rank value；Initialize ListexecutionFor sky, for depositing
Put the task execution sequence being subsequently generated；
Step 3: judging whether queue GQueue is sky, if queue not empty, goes to step (4)；Go to step (6)；
Step 4: from task ttall(i.e. tail task) starts, and passes through R (ti)=Mi+max{N(h,i),th∈tpre}+max{N(i,j)+R
(tj),tj∈tsucSuccessively calculate the rank value of each task upwards step by step；
Step 5: the rank value of calculating is stored in ListRIn, it deletes rank value in GQueue and calculates the task node completed, go to step
(3)；
Step 6: traversal ListRIn all rank value, carry out descending arrangement, determine task priority；
Step 7: by ListRIn ready task according to priority be successively stored in ListexecutionIn, and in ListR
Middle deletion；
Step 8: initialization Makespan is sky, is used to store each workflow deadline sequence；
Step 9: judging ListexecutionWhether it is sky, if sequence non-empty, goes to step (10)；If sequence is empty, go to step
(17)；
Step 10: judging the corresponding relationship of predecessor task and subsequent tasks, server is selected, if predecessor task and subsequent tasks
When same server executes, without considering data communication time, if one-to-one situation, (11) are gone to step；If one-to-many
Situation goes to step (12)；If many-one situation goes to step (13)；
Step 11: server where subsequent tasks selection predecessor task can be sought under the premise of not interfering high-priority task
The free time section of server is looked for, segmentation executes, and goes to step (14)；
Step 12: multiple subsequent tasks successively select server according to priority, select deadline earliest server with robbing
The method for accounting for formula executes, and when identical there are the deadline of multiple servers, selects the server of Starting Executing Time the latest,
Go to step (14)；
Step 13: subsequent tasks Starting Executing Time depends on the predecessor task completed the latest, selects the clothes that the deadline is earliest
Business device selects the server of Starting Executing Time the latest, goes to step when identical there are the deadline of multiple servers
(14)；
Step 14: selected server is judged after previous task execution, before current task executes, if deposit during idle time
Between, if it does, going to step (15)；If it does not, going to step (16)；
Step 15: the dependence of previous task Yu its subsequent tasks is further judged, if complete in previous task execution
At before starting later with its subsequent tasks, there are one section of free times, and free time section occupies the clothes without other tasks
Business device, then the free time section to server carries out dynamic voltage frequency adjustment；Otherwise, (16) are gone to step；
Step 16: the deadline of each task being recorded in sequence Makespan, and from ListexecutionMiddle deletion is completed
Task, go to step (9)；
Step 17: output task completion time sequence table.
The Work flow model is to carry out applying modeling, each node on behalf in Work flow model using oriented no circulation figure
One task, information includes task names in node, selects the processing time of the server executed and selected server；Only
After having all father's tasks of subtask all to execute, subtask can just be handled；Before directed edge represents between node
After constrain, weight represents task in the data communication time of non-same server.
The workflow schedule method that the use concurrently executes Optimized model is divided into four-stage: task order calculates rank
Section, task pool construct stage, resource selection stage, dynamic energy-saving adjustment.
The task order calculation stages, comprehensively consider according to the average communication data of the subsequent grade of task and forerunner and set
Task order is set, and is more than the grade for considering subsequent tasks, promotes to distribute to as far as possible there are the task of dependence same
Equipment executes, to reduce scheduling length, achievees the effect that shorten task completion time.
The resource selection stage successively selects resource according to the execution sequence of ready task in task pool, rank value compared with
High task priority is higher than the lesser task of rank value and carries out classification consideration for different resource selection situations.
The utility model has the advantages that
1. of the invention consider low in raising resource utilization simultaneously while shortening task execution time and reducing server execution
The problem of energy consumption: previous Multi-workflow concurrent scheduling research only considers to shorten the execution time of workflow.The present invention is using concurrent
Execute Optimized model.1. using the calculative strategy of task order, reduce scheduling length, reaches the effect for shortening task completion time
Fruit；2. ensure that the fairness in scheduling implementation procedure using the strategy of building task pool；3. being pressed using the task in task pool
Priority reduces task completion time, improves the utilization rate of resource by resource selection strategy；4. passing through dynamic electric voltage-frequency
The mode of rate adjustment, reaching reduces the target for executing energy consumption.
The present invention innovates the calculation method of task order on the basis of HEFT, by the grade of subsequent tasks and
Average communication data between predecessor task takes into account as the parameter of calculating task order, makes the task there are dependence
Same server is distributed to as far as possible to execute, and is reduced task schedule length, is reached the requirement of the deadline of shortening task；Together
When, propose new resource selection standard, by the corresponding relationship of forerunner and subsequent tasks, be divided into it is one-to-one, it is one-to-many and
The case where multi-to-multi, different situations correspond to different selection criteria, and the resource of selection is made to be more in line with actual conditions, reduce service
The waiting of device and idle situation meet the demand for improving resource utilization.
Detailed description of the invention:
Attached drawing 1 is flow chart of the invention.
Attached drawing 2 is task illustrated example of the invention.
Attached drawing 3 is calculating time of the present invention to each task node of attached drawing 2.
Attached drawing 4 is the present invention for the exemplary scheduling result Gantt chart of task image shown in attached drawing 2.
Attached drawing 5 is the prior art to the exemplary scheduling result Gantt chart of task image shown in attached drawing 2.
Attached drawing 6 is the prior art and the present invention to the exemplary task rank value comparison diagram of task image shown in attached drawing 2.
Specific embodiment:
Embodiment 1:
A kind of more scientific workflows based on cloud environment concurrently execute dispatching method, and this method comprises the following steps:
A kind of more scientific workflows based on cloud environment concurrently execute dispatching method, it is characterized in that: concurrently executing Optimized model.It is first
First, using the calculative strategy of task order, reduce scheduling length, achieve the effect that shorten task completion time；Then, using structure
The strategy for building task pool ensure that the fairness in scheduling implementation procedure；Secondly, according to priority being led to using the task in task pool
Resource selection strategy is crossed, task completion time is reduced, improves the utilization rate of resource；Finally, passing through dynamic voltage frequency tune
Whole mode, reaching reduces the target for executing energy consumption.
Embodiment 2:
More scientific workflows based on cloud environment concurrently execute dispatching method, and the dispatching method specific implementation step is as follows:
Step 1: each task node information of each scientific workflow being stored in one and is initialized as in empty queue GQueue；
Step 2: initialization ListRFor sky, it is used to store the collating sequence of rank value；Initialize ListexecutionFor sky, for depositing
Put the task execution sequence being subsequently generated；
Step 3: judging whether queue GQueue is sky, if queue not empty, goes to step (4)；Go to step (6)；
Step 4: from task ttall(i.e. tail task) starts, and passes through R (ti)=Mi+max{N(h,i),th∈tpre}+max{N(i,j)+R
(tj),tj∈tsucSuccessively calculate the rank value of each task upwards step by step；
Step 5: the rank value of calculating is stored in ListRIn, it deletes rank value in GQueue and calculates the task node completed, go to step
(3)；
Step 6: traversal ListRIn all rank value, carry out descending arrangement, determine task priority；
Step 7: by ListRIn ready task according to priority be successively stored in ListexecutionIn, and in ListR
Middle deletion；
Step 8: initialization Makespan is sky, is used to store each workflow deadline sequence；
Step 9: judging ListexecutionWhether it is sky, if sequence non-empty, goes to step (10)；If sequence is empty, go to step
(17)；
Step 10: judging the corresponding relationship of predecessor task and subsequent tasks, server is selected, if predecessor task and subsequent tasks
When same server executes, without considering data communication time, if one-to-one situation, (11) are gone to step；If one-to-many
Situation goes to step (12)；If many-one situation goes to step (13)；
Step 11: server where subsequent tasks selection predecessor task can be sought under the premise of not interfering high-priority task
The free time section of server is looked for, segmentation executes, and goes to step (14)；
Step 12: multiple subsequent tasks successively select server according to priority, select deadline earliest server with robbing
The method for accounting for formula executes, and when identical there are the deadline of multiple servers, selects the server of Starting Executing Time the latest,
Go to step (14)；
Step 13: subsequent tasks Starting Executing Time depends on the predecessor task completed the latest, selects the clothes that the deadline is earliest
Business device selects the server of Starting Executing Time the latest, goes to step when identical there are the deadline of multiple servers
(14)；
Step 14: selected server is judged after previous task execution, before current task executes, if deposit during idle time
Between, if it does, going to step (15)；If it does not, going to step (16)；
Step 15: the dependence of previous task Yu its subsequent tasks is further judged, if complete in previous task execution
At before starting later with its subsequent tasks, there are one section of free times, and free time section occupies the clothes without other tasks
Business device, then the free time section to server carries out dynamic voltage frequency adjustment；Otherwise, (16) are gone to step；
Step 16: the deadline of each task being recorded in sequence Makespan, and from ListexecutionMiddle deletion is completed
Task, go to step (9)；
Step 17: output task completion time sequence table.
Embodiment 3:
The Work flow model is to carry out applying modeling, each node on behalf one in Work flow model using oriented no circulation figure
Task, information includes task names in node, selects the processing time of the server executed and selected server；It is only sub
After all father's tasks of task all execute, subtask can just be handled；Directed edge represents the front and back between node about
Beam, weight represent task in the data communication time of non-same server.
The workflow schedule method that the use concurrently executes Optimized model is divided into four-stage: task order calculates rank
Section, task pool construct stage, resource selection stage, dynamic energy-saving adjustment.
The task order calculation stages, comprehensively consider according to the average communication data of the subsequent grade of task and forerunner and set
Task order is set, and is more than the grade for considering subsequent tasks, promotes to distribute to as far as possible there are the task of dependence same
Equipment executes, to reduce scheduling length, achievees the effect that shorten task completion time.
The resource selection stage successively selects resource according to the execution sequence of ready task in task pool, rank value compared with
High task priority is higher than the lesser task of rank value and carries out classification consideration for different resource selection situations.
Embodiment 4:
Above-mentioned more scientific workflows based on cloud environment concurrently execute dispatching method, task execution sequence generating module:
Since the tail task in DAG model in an iterative manner, from the bottom up in step-by-step calculation workflow each task order
Value, wherein the calculating parameter of rank value includes: the average calculation times of task, the maximum value with predecessor task average communication data,
And the maximum value with the sum of subsequent tasks average communication data and subsequent tasks rank value, make that there are the task of dependence is most
Same server may be distributed to, scheduling length is reduced, shortens task completion time.Resulting rank value will be calculated to be stored in
ListRIn, after the rank value of tasks all in workflow calculates, by ListRIn all rank value carry out descending arrangements, order
Value is bigger, and the priority for representing task is higher.Ready task is selected according to priority orders from each workflow to be put into
ListexecutionIn, it is built into task pool, promotes the fairness in scheduling process.
Embodiment 5:
Above-mentioned more scientific workflows based on cloud environment concurrently execute dispatching method, resource selection module:
First, it is determined that the corresponding relationship of predecessor task and subsequent tasks, is divided into one-to-one, one-to-many and multi-to-multi relationship, no
Different resource selection strategies is corresponded to corresponding relationship；If predecessor task and subsequent tasks are when same server executes, nothing
It need to consider data communication time；In one-to-one situation, subsequent tasks select the server where predecessor task, are not interfering height
Under the premise of priority tasks execute, the free time section of server can be found, segmentation executes；It is more in one-to-many situation
A subsequent tasks successively select deadline earliest server according to priority, are executed using preemptive type mode；Such as task A
Same server is selected with B, A priority is high but data communication time is long, and B priority is low but data communication time is short, reaches in A
Before arriving, B can be first carried out, and after A reaches, A is executed, and B temporary suspension waits, and after the completion of waiting A, B is continued to execute；Multipair
In the case where one, the server that the deadline is earliest is selected, when identical there are the deadline of multiple servers, selection starts
Execute the server of time the latest；The time that subsequent tasks start depends on the predecessor task completed the latest；For different situations
Difference is treated, and the server for selecting the deadline earliest executes, and is reduced task completion time, is improved the utilization rate of resource.
Embodiment 6:
Above-mentioned more scientific workflows based on cloud environment concurrently execute dispatching method, and dynamic energy-saving adjusts module:
Detailed judgement is carried out to the free time of server, it is determined whether dynamic energy-saving adjustment can be carried out.If task C is selected
Server 1 is selected, the previous task executed on the server is A, and the subsequent tasks of task A are B；So, A is first determined whether
It executes before terminating to execute with C, this period, whether server 1 was idle, the further judgement if idle；Judge the resource of B
Distribution and executive condition, if before A starts execution with B after executing, there are free time sections, and in the period
In, server 1 is occupied without other tasks, it is possible to carry out dynamic voltage frequency adjustment, drop to the period of server 1
Low running frequency reduces system energy consumption.
Embodiment 7:
Above-mentioned more scientific workflows based on cloud environment concurrently execute dispatching method, more sections of this technology mainly for cloud environment
Workflow schedule task is learned, generallys use directed acyclic graph to indicate, regulation goal, which is in the case where meeting constraint condition, is
Workflow determines position in Gantt chart；As shown in Fig. 2, information includes task names in node, selects the server executed,
And the processing time of selected server；After only all father's tasks of subtask all execute, subtask just can be into
Row processing；Directed edge represents the precedence constraints between node, and weight represents task in the data communication time of non-same server.
Entire task image is made of the descendant nodes of root node and root node；In scheduling, the finger on side in directed acyclic graph
To meaning: being pointed to node is the next task of node of setting out, and is also father node task；Otherwise it is considered that set out node be by
It is directed toward the tight preceding task of node, is also child's task.
Embodiment 8:
Above-mentioned more scientific workflows based on cloud environment concurrently execute dispatching method, as shown in Fig. 2, Work flow model example
Figure is made by 10 tasks, and t1, t2, t4 task have multiple subsequent tasks to belong to one-to-many situation；T8, t9, t10 task have multiple
Predecessor task belongs to many-one situation；When there are the tasks of dependence when different server executes, need to carry out data
Communication, takes some time；As shown in Fig. 3, each task in workflow can choose different servers and execute, no
The processing time that same server is wanted for different required by task is also different；Therefore, when task is when selecting server,
It not only needs to consider data communication time, it is also contemplated that the processing time of different server, comprehensively considers selection and pay a price most
Few server executes.
With reference to the accompanying drawings 1 pair of attached drawing 2, attached drawing 3 workflow be scheduled and draw Gantt chart, it is each as shown in Fig. 4 to appoint
It is engaged in selected execute server and process time, the workflow schedule deadline is 76.
Embodiment 9:
Above-mentioned more scientific workflows based on cloud environment concurrently execute dispatching method, examples comparative:
Using the critical path (CPOP) in existing more mode of priority isomery earliest finish time (HEFT) method and processor
Method is respectively scheduled the task of attached drawing 2, attached drawing 3, obtains Gantt chart such as attached drawing 5；Existing method and this method respectively into
Row rank value calculates as shown in Fig. 6, and the priority of task choosing resource is determined by rank value, by comparison, it was found that, this method is excellent
First grade determine it is more reasonable, on follow-up work influence deeper task can selection resource earlier execute.
Therefore, the present invention is completely new method, concurrently executes scheduler task for more scientific workflows under cloud environment.
Claims (6)
1. a kind of more scientific workflows based on cloud environment concurrently execute dispatching method, it is characterized in that: using optimization is concurrently executed
Model；
1. using the calculative strategy of task order, scheduling length is reduced, achievees the effect that shorten task completion time；2. using building
The strategy of task pool ensure that the fairness in scheduling implementation procedure；Money is according to priority passed through using the task in task pool
Source selection strategy reduces task completion time, improves the utilization rate of resource；Dynamic voltage frequency adjust by way of pair
There are the resources of free time section to carry out voltage adjustment, and reaching reduces the target for executing energy consumption.
2. more scientific workflows according to claim 1 based on cloud environment concurrently execute dispatching method, the scheduling
The specific implementation step of method is as follows:
Step 1: each task node information of each scientific workflow being stored in one and is initialized as in empty queue GQueue；
Step 2: initialization ListRFor sky, it is used to store the collating sequence of rank value；Initialize ListexecutionFor sky, it is used to store
The task execution sequence being subsequently generated;
Step 3: judging whether queue GQueue is sky, if queue not empty, goes to step (4)；Go to step (6);
Step 4: from task ttall(i.e. tail task) starts, and passes through R (ti)=Mi+max{N(h,i),th∈tpre}+max{N(i,j)+R
(tj),tj∈tsucSuccessively calculate the rank value of each task upwards step by step；
Step 5: the rank value of calculating is stored in ListRIn, it deletes rank value in GQueue and calculates the task node completed, go to step
(3);
Step 6: traversal ListRIn all rank value, carry out descending arrangement, determine task priority;
Step 7: by ListRIn ready task according to priority be successively stored in ListexecutionIn, and in ListR
Middle deletion;
Step 8: initialization Makespan is sky, is used to store each workflow deadline sequence;
Step 9: judging ListexecutionWhether it is sky, if sequence non-empty, goes to step (10)；If sequence is empty, go to step
(17);
Step 10: judging the corresponding relationship of predecessor task and subsequent tasks, server is selected, if predecessor task and subsequent tasks
When same server executes, without considering data communication time, if one-to-one situation, (11) are gone to step；If one-to-many
Situation goes to step (12)；If many-one situation goes to step (13)；
Step 11: server where subsequent tasks selection predecessor task can be sought under the premise of not interfering high-priority task
The free time section of server is looked for, segmentation executes, and goes to step (14);
Step 12: multiple subsequent tasks successively select server according to priority, select deadline earliest server with robbing
The method for accounting for formula executes, and when identical there are the deadline of multiple servers, selects the server of Starting Executing Time the latest,
Go to step (14)；
Step 13: subsequent tasks Starting Executing Time depends on the predecessor task completed the latest, selects the clothes that the deadline is earliest
Business device selects the server of Starting Executing Time the latest, goes to step when identical there are the deadline of multiple servers
(14)；
Step 14: selected server is judged after previous task execution, before current task executes, if deposit during idle time
Between, if it does, going to step (15)；If it does not, going to step (16)；
Step 15: the dependence of previous task Yu its subsequent tasks is further judged, if complete in previous task execution
At before starting later with its subsequent tasks, there are one section of free times, and free time section occupies the clothes without other tasks
Business device, then the free time section to server carries out dynamic voltage frequency adjustment；Otherwise, (16) are gone to step；
Step 16: the deadline of each task being recorded in sequence Makespan, and from ListexecutionMiddle deletion is completed
Task, go to step (9)；
Step 17: output task completion time sequence table.
3. more scientific workflows of the consideration based on cloud environment according to claim 1 or 2 concurrently execute dispatching method, special
Sign is: the Work flow model is to carry out applying modeling, each node on behalf one in Work flow model using oriented no circulation figure
A task, information includes task names in node, selects the processing time of the server executed and selected server；Only
After all father's tasks of subtask all execute, subtask can just be handled；Directed edge represents the front and back between node
Constraint, weight represent task in the data communication time of non-same server.
4. more scientific workflows of the consideration based on cloud environment according to claim 1 or 2 or 3 concurrently execute dispatching method,
It is characterized in that: the workflow schedule method that the use concurrently executes Optimized model is divided into four-stage: task order calculates rank
Section, task pool construct stage, resource selection stage, dynamic energy-saving adjustment.
5. more scientific workflows of the consideration based on cloud environment according to claim 1 or 2 or 3 or 4 concurrently execute dispatching party
Method, it is characterized in that: the task order calculation stages, are examined according to the average communication data of the subsequent grade of task and forerunner synthesis
Consider setting task order, and be more than and consider the grades of subsequent tasks, promotes that there are the tasks of dependence to distribute to as far as possible
Same equipment executes, to reduce scheduling length, achievees the effect that shorten task completion time.
6. consideration described according to claim 1 or 2 or 3 or 4 or 5 concurrently executes scheduling based on more scientific workflows of cloud environment
Method, it is characterized in that: the resource selection stage, successively selects resource according to the execution sequence of ready task in task pool,
The higher task priority of rank value is higher than the lesser task of rank value and carries out classification consideration for different resource selection situations.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910557811.4A CN110264097A (en) | 2019-06-26 | 2019-06-26 | More scientific workflows based on cloud environment concurrently execute dispatching method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910557811.4A CN110264097A (en) | 2019-06-26 | 2019-06-26 | More scientific workflows based on cloud environment concurrently execute dispatching method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110264097A true CN110264097A (en) | 2019-09-20 |
Family
ID=67921541
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910557811.4A Pending CN110264097A (en) | 2019-06-26 | 2019-06-26 | More scientific workflows based on cloud environment concurrently execute dispatching method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110264097A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111756653A (en) * | 2020-06-04 | 2020-10-09 | 北京理工大学 | Multi-coflow scheduling method based on deep reinforcement learning of graph neural network |
| CN113064710A (en) * | 2021-04-15 | 2021-07-02 | 北京理工大学 | Cloud workflow scheduling method and system |
| CN113485819A (en) * | 2021-08-03 | 2021-10-08 | 北京八分量信息科技有限公司 | Heterogeneous task preprocessing method and device and related products |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102594433A (en) * | 2011-12-14 | 2012-07-18 | 大连大学 | Dynamic priority dispatching method of satellite network many-to-one communication conflicts |
| CN106648831A (en) * | 2016-12-08 | 2017-05-10 | 杭州电子科技大学 | Cloud workflow scheduling method based on firefly algorithm and dynamic priority algorithm |
| CN107168770A (en) * | 2017-04-14 | 2017-09-15 | 中国人民解放军国防科学技术大学 | A kind of cloud data center workflow schedule of low energy consumption and resource provision method |
| CN107656799A (en) * | 2017-11-06 | 2018-02-02 | 福建师范大学 | The workflow schedule method of communication and calculation cost is considered under a kind of more cloud environments |
| US20180285146A1 (en) * | 2017-03-31 | 2018-10-04 | International Business Machines Corporation | Workflow handling in a multi-tenant cloud environment |
| CN108647084A (en) * | 2018-05-08 | 2018-10-12 | 武汉轻工大学 | Efficiency cloud method for scheduling task |
| CN109032756A (en) * | 2018-07-04 | 2018-12-18 | 中国人民解放军国防科技大学 | Scheduling method of virtualized cloud data center |
| CN109359888A (en) * | 2018-11-15 | 2019-02-19 | 哈尔滨理工大学 | There is the integrated dispatch method of close linking constraint between a kind of more equipment processes |
| CN109730701A (en) * | 2019-01-03 | 2019-05-10 | 中国电子科技集团公司电子科学研究院 | A kind of acquisition methods and device of mood data |
-
2019
- 2019-06-26 CN CN201910557811.4A patent/CN110264097A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102594433A (en) * | 2011-12-14 | 2012-07-18 | 大连大学 | Dynamic priority dispatching method of satellite network many-to-one communication conflicts |
| CN106648831A (en) * | 2016-12-08 | 2017-05-10 | 杭州电子科技大学 | Cloud workflow scheduling method based on firefly algorithm and dynamic priority algorithm |
| US20180285146A1 (en) * | 2017-03-31 | 2018-10-04 | International Business Machines Corporation | Workflow handling in a multi-tenant cloud environment |
| CN107168770A (en) * | 2017-04-14 | 2017-09-15 | 中国人民解放军国防科学技术大学 | A kind of cloud data center workflow schedule of low energy consumption and resource provision method |
| CN107656799A (en) * | 2017-11-06 | 2018-02-02 | 福建师范大学 | The workflow schedule method of communication and calculation cost is considered under a kind of more cloud environments |
| CN108647084A (en) * | 2018-05-08 | 2018-10-12 | 武汉轻工大学 | Efficiency cloud method for scheduling task |
| CN109032756A (en) * | 2018-07-04 | 2018-12-18 | 中国人民解放军国防科技大学 | Scheduling method of virtualized cloud data center |
| CN109359888A (en) * | 2018-11-15 | 2019-02-19 | 哈尔滨理工大学 | There is the integrated dispatch method of close linking constraint between a kind of more equipment processes |
| CN109730701A (en) * | 2019-01-03 | 2019-05-10 | 中国电子科技集团公司电子科学研究院 | A kind of acquisition methods and device of mood data |
Non-Patent Citations (2)
| Title |
|---|
| 郑宏升: "《基于萤火虫算法和动态优先级的多QoS云工作流调度》", 《计算机集成制造系统》 * |
| 陈彦橦: "《云科学工作流截止期限约束代价优化调度算法》", 《计算机科学与探索》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111756653A (en) * | 2020-06-04 | 2020-10-09 | 北京理工大学 | Multi-coflow scheduling method based on deep reinforcement learning of graph neural network |
| CN113064710A (en) * | 2021-04-15 | 2021-07-02 | 北京理工大学 | Cloud workflow scheduling method and system |
| CN113485819A (en) * | 2021-08-03 | 2021-10-08 | 北京八分量信息科技有限公司 | Heterogeneous task preprocessing method and device and related products |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101237469B (en) | Method for optimizing multi-QoS grid workflow based on ant group algorithm | |
| CN108345501B (en) | Distributed resource scheduling method and system | |
| CN110264097A (en) | More scientific workflows based on cloud environment concurrently execute dispatching method | |
| CN107992359B (en) | Task scheduling method for cost perception in cloud environment | |
| CN110308967B (en) | Workflow cost-delay optimization task allocation method based on hybrid cloud | |
| CN104168318A (en) | Resource service system and resource distribution method thereof | |
| CN112416585B (en) | Deep learning-oriented GPU resource management and intelligent scheduling method | |
| CN105389206B (en) | A kind of cloud computation data center resources of virtual machine quickly configuration method | |
| CN109582448A (en) | A kind of edge calculations method for scheduling task towards criticality and timeliness | |
| CN109948848A (en) | Research-on-research flows down the Cost Optimization dispatching method of deadline constraint in a kind of cloud | |
| Dittakavi | An extensive exploration of techniques for resource and cost management in contemporary cloud computing environments | |
| CN108509603B (en) | A kind of adaptive dynamic dispatching method and system of data warehouse | |
| CN106934539A (en) | It is a kind of with limited and expense restriction workflow schedule method | |
| CN110908772A (en) | Energy-saving scheduling method for improving reliability of multiple workflows | |
| US8281313B1 (en) | Scheduling computer processing jobs that have stages and precedence constraints among the stages | |
| CN116467076A (en) | Multi-cluster scheduling method and system based on cluster available resources | |
| CN115564242A (en) | Ship power equipment-oriented scheduling method and system for preemptible task maintenance personnel | |
| CN108647084A (en) | Efficiency cloud method for scheduling task | |
| Singh et al. | A comparative study of various scheduling algorithms in cloud computing | |
| CN106951313A (en) | The sub- time limit acquisition methods of Multi-workflow shared resource cooperative scheduling | |
| Li et al. | A QoS-based scheduling algorithm for instance-intensive workflows in cloud environment | |
| CN112306642A (en) | Workflow scheduling method based on stable matching game theory | |
| CN110084507A (en) | The scientific workflow method for optimizing scheduling of perception is classified under cloud computing environment | |
| CN110928648A (en) | Heuristic and intelligent computing-fused cloud workflow segmentation online scheduling optimization method | |
| CN109343940A (en) | Multimedia Task method for optimizing scheduling in a kind of cloud platform |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190920 |