CN105260230A - Resource scheduling method for data center virtual machine based on segmented service level agreement - Google Patents
Resource scheduling method for data center virtual machine based on segmented service level agreement Download PDFInfo
- Publication number
- CN105260230A CN105260230A CN201510733449.3A CN201510733449A CN105260230A CN 105260230 A CN105260230 A CN 105260230A CN 201510733449 A CN201510733449 A CN 201510733449A CN 105260230 A CN105260230 A CN 105260230A
- Authority
- CN
- China
- Prior art keywords
- virtual machine
- resources
- level agreement
- service
- data center
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a resource scheduling method for a data center virtual machine based on a segmented service level agreement. According to the method, operation of a user as well as execution time of each stage is divided into three parts according to the execution response time and the execution process of the operation of the user in a cloud environment, wherein the three parts are as follows: 1, correspondingly segmenting a service level agreement according to the execution process of the operation, defining effective unit time cost UUTC, taking resource of the virtual machine as state space for reinforced learning, and taking distribution and recycle as the motion space of the reinforced learning; 2, utilizing the reinforced learning for resource scheduling policy leaning and making a rule; 3, conducting execution from the step one on follow-up arrived cloud operation streams. The method has the advantages that under the premise that the service level agreement is ensured, consumption of resources such as cpu, internal memory and bandwidth is reduced; otherwise, consumption of the resources is enlarged.
Description
Technical field
The invention belongs to cloud platform technology field, particularly relate to a kind of data center's resources of virtual machine dispatching method based on segmentation service-level agreement.
Background technology
Under dynamic cloud computing environment, resources of virtual machine and the online dynamic optimization of application system parameter configure very difficult.On the one hand, in cloud computing, the scale of all kinds of bottom hardware resource is normally very huge.On the other hand, dynamic cloud computing environment has very large uncertainty.Therefore, dynamic cloud computing environment objectively requires the online dynamic optimization configuration being realized resources of virtual machine and application system parameter by height adaptive means.In addition, in cloud computing, the configuration of resources of virtual machine and application system parameter often interacts, and needs to coordinate configuration, and adjustment just may not can improve resource utilization and application service performance on the one hand separately.Dynamic cloud computing environment objectively also requires that resources of virtual machine and application system parameter are carried out online dynamic self-adapting with cooperative mode and distributed rationally.
In recent years, research focuses in cloud platform on virtual machine Placement Problems by domestic and international many scholars, and suitable virtual machine placement method can improve the utilization factor of resource, and the reliability of elevator system and Consumer's Experience, economize on resources.Model for Multi-Objective Optimization is introduced in document " Amatrixtransformationalgorithmforvirtualmachineplacement incloud ", propose a kind of algorithm based on matrixing, effectively control required physical machine quantity of finishing the work, improve the utilization factor of resource simultaneously.But the Optimized model of this algorithm has only paid close attention to the idle situation of multi dimensional resource in whole resource pool, and do not consider the utilization factor situation of resource in single physical machine, the non-load balanced case of whole algorithm to each physics does not take in, and thus data center's operational reliability can reduce.Document " the virtual machine Placement Strategy based on particle swarm optimization algorithm " proposes a kind of virtual machine Placement Strategy based on particle cluster algorithm, effectively improves user task request response time, improves the task processing speed of whole system.But this algorithm single using CPU to the standard of data processing delay as solved function optimum solution, do not consider the problem of load balancing of system.In document " Adynamicpriorityschedulingalgorithmonservicerequestsched ulingincloudcomputing ", system resource is monitored in real time, and calculate Current resource utilization factor, user task is assigned in the minimum physical machine of resource utilization.This algorithm can provide a good QoS for user, but resource utilization ratio is not high.Document " Amulti-objectiveantcolonysystemalgorithmforvirtualmachin eplacementincloudcomputing " proposes the physical machine load-balancing method based on ant group algorithm, this algorithm adopts didactic algorithm realization complexity very high, affect the stand-by period of user, algorithm the convergence speed is excessively slow.Document " Sharesandutilitiesbasedpowerconsolidationinvirtualizedse rverenvironments " proposes a kind of resources of virtual machine allocation algorithm based on distributing maximum resource, least resource and shared resource characteristic.But this algorithm carries out Resourse Distribute mainly for single physical machine, the not good resource to other physical machine in cloud system carries out reasonable distribution, is not suitable for data center's colony dispatching environment.
Summary of the invention
The object of the present invention is to provide a kind of data center's resources of virtual machine dispatching method based on segmentation service-level agreement, be intended under Deterministic service level agreements prerequisite, improve resource utilization to greatest extent in dynamic cloud computing environment.
The present invention is achieved in that a kind of data center's resources of virtual machine dispatching method based on segmentation service-level agreement comprises:
Step one, perform response time and implementation according to cloud environment user job, by user job and the execution time in each stage, be divided into three parts, be respectively cloud job queue time JQT, cloud Job execution time JET and cloud job transfer time JTT;
Step 2, according to Job execution process, service-level agreement correspondence is carried out segmentation;
Step 3, definition effective unit time cost UUTC are
the expense that Totalcost pays for executing a user job, concrete value is consulted by cloud service provider and user; T
totfor the operation response time, its value is T
tot=JQT+JET+JTT;
Step 4, with resources of virtual machine, as the state space of intensified learning; To distribute and to reclaim as intensified learning motion space;
Step 5, utilize intensified learning to carry out the study of resources of virtual machine scheduling strategy, lay down a regulation;
Step 6, cloud job stream for follow-up arrival, perform at the beginning from step.
Further, in step 2, service-level agreement correspondence is carried out segmentation, meets following formula respectively:
JQT≤SLA
JQT；
JET≤SLA
JET；
JTT≤SLA
JTT。
Further, in step 4, to distribute and to reclaim as intensified learning motion space, Reward Program is described below immediately:
(1) if the UUTC of current work is greater than average UUTC, and this operation meets service-level agreement and QoS constraint, then return is 1;
(2) if current work does not meet service-level agreement and QoS constraint, then return is for-1;
(3) other situations, return is 0.
Further, the rule formulated in step 5 is:
(1) if operation certain stage in the process of implementation, violate the constraint of segmentation service-level agreement, then this operation is in follow-up implementation, increases the resources of virtual machine distributed;
(2) if the UUTC of this operation is less than average UUTC, then this operation is in follow-up implementation, subtracts under absorbed resources of virtual machine;
Wherein, cpu resource increases at every turn or reduces 1, and internal memory increases at every turn or reduces 256M, and bandwidth increases at every turn or reduces 256kbps.
The virtual function of data center of the present invention is according to the rate that the reaches dynamic conditioning resources of virtual machine of user's submit job.Such as, when the rate that reaches of user job is lower, the present invention, under the prerequisite guaranteeing service quality rating (SLA), reduces the resources such as cpu, internal memory, bandwidth; Otherwise, then corresponding resource is increased.
The Performance comparision of the inventive method and other similar approach
Accompanying drawing explanation
Fig. 1 is the data center's resources of virtual machine dispatching method process flow diagram based on segmentation service-level agreement that the embodiment of the present invention provides.
Fig. 2 is under the different pressures that provides of the embodiment of the present invention, and what the present invention and utilization factor method used CPU number compares schematic diagram.
Fig. 3 be the present invention of providing of the embodiment of the present invention with utilization factor resource allocation methods and original Q resource allocation methods compare schematic diagram.
Embodiment
In order to make object of the present invention, technical scheme and advantage clearly understand, below in conjunction with embodiment, the present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explain the present invention, be not intended to limit the present invention.
Below in conjunction with drawings and the specific embodiments, application principle of the present invention is further described.
As shown in Figure 1, the data center's resources of virtual machine dispatching method based on segmentation service-level agreement of the embodiment of the present invention comprises:
S101, perform response time and implementation according to cloud environment user job, by user job and the execution time in each stage, be divided into three parts, be respectively cloud job queue time JQT, cloud Job execution time JET and cloud job transfer time JTT;
S102, according to Job execution process, service-level agreement correspondence is carried out segmentation;
S103, definition effective unit time cost UUTC;
S104, with resources of virtual machine, as the state space of intensified learning; To distribute and to reclaim as intensified learning motion space;
S105, utilize intensified learning to carry out the study of resources of virtual machine scheduling strategy, lay down a regulation;
S106, cloud job stream for follow-up arrival, perform from step S101.
Further,
the expense that Totalcost pays for executing a user job, concrete value is consulted by cloud service provider and user; T
totfor the operation response time, its value is T
tot=JQT+JET+JTT.
Further, in step S102, service-level agreement correspondence is carried out segmentation, meets following formula respectively:
JQT≤SLA
JQT；
JET≤SLA
JET；
JTT≤SLA
JTT。
Further, in step S104, to distribute and to reclaim as intensified learning motion space, Reward Program is described below immediately:
(1) if the UUTC of current work is greater than average UUTC, and this operation meets service-level agreement and QoS constraint, then return is 1;
(2) if current work does not meet service-level agreement and QoS constraint, then return is for-1;
(3) other situations, return is 0.
Described resources of virtual machine comprises cpu, internal memory and bandwidth, and the state-space representation of each virtual machine is a vector, value can not exceed virtual machine have the upper limit of resource.Suppose that a physical machine has 8 cpu, 8G internal memories and 100M bandwidth, then its virtual after the state space of certain virtual machine can be expressed as (1,2,2), implication has 1 cpu for this virtual machine, the bandwidth of 2G internal memory and 2M.
For i-th virtual machine, to the resources of virtual machine that it has, possible action comprises increase resource, and resource is constant and reduce resource, and these three kinds of actions are expressed as 1,0 and-1.Suppose that the state space of certain virtual machine is for (1,2,2), then the motion space of making during certain decision-making is (0,1 ,-1), and it remains unchanged containing promising cpu resource, and internal memory increases 512M, and bandwidth reduces 0.5M.
Further, the rule formulated in step S105 is:
(1) if operation certain stage in the process of implementation, violate the constraint of segmentation service-level agreement, then this operation is in follow-up implementation, increases the resources of virtual machine distributed;
(2) if the UUTC of this operation is less than average UUTC, then this operation is in follow-up implementation, subtracts under absorbed resources of virtual machine;
Wherein, cpu resource increases at every turn or reduces 1, and internal memory increases at every turn or reduces 256M, and bandwidth increases at every turn or reduces 256kbps.
Below in conjunction with experiment, effect of the present invention is further described.
SPECjbb2005 platform is utilized to verify the performance of this invention further, SPECjbb2005 benchmark test simulates a three-tier architecture environment to carry out the test of JAVA application server performance, under testing each warehouses respectively in experiment, in experimentation, run utilization factor resource regulating method and the inventive method 10 times respectively, average experiment result as shown in Figure 2.Fig. 3 is the present invention and the comparing of utilization factor resource allocation methods and original Q resource allocation methods.
The result (Bops) on SPECjbb2005 platform
| Warehouse | Maximum | Utilization factor resource regulating method | The inventive method |
| 1 | 32103.43 | 32847.88 | 33767.96 |
| 2 | 63252.47 | 59510.64 | 59951.13 |
| 3 | 87267.56 | 82380.99 | 83919.59 |
| 4 | 105719.64 | 104081.37 | 99212.12 |
| 5 | 115648.76 | 117200.12 | 116671.41 |
| 6 | 119057.39 | 121530.59 | 122071.97 |
| 7 | 123759.19 | 126187.14 | 124289.81 |
| 8 | 120844.70 | 122457.18 | 122758.30 |
Bops completes how many JAVA business operations (BusinessOperationPerSecond) p.s..
The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, all any amendments done within the spirit and principles in the present invention, equivalent replacement and improvement etc., all should be included within protection scope of the present invention.
Claims (5)
1. based on data center's resources of virtual machine dispatching method of segmentation service-level agreement, it is characterized in that, the described data center's resources of virtual machine dispatching method based on segmentation service-level agreement comprises:
Perform response time and implementation according to cloud environment user job, by user job and the execution time in each stage, be divided into three parts, be respectively cloud job queue time JQT, cloud Job execution time JET and cloud job transfer time JTT;
According to Job execution process, service-level agreement correspondence is carried out segmentation;
Definition effective unit time cost UUTC is
the expense that Totalcost pays for executing a user job, concrete value is consulted by cloud service provider and user; T
totfor the operation response time, its value is T
tot=JQT+JET+JTT;
With resources of virtual machine, as the state space of intensified learning; To distribute and to reclaim as intensified learning motion space;
Utilize intensified learning to carry out the study of resources of virtual machine scheduling strategy, lay down a regulation;
2., as claimed in claim 1 based on data center's resources of virtual machine dispatching method of segmentation service-level agreement, it is characterized in that, service-level agreement correspondence is carried out segmentation, meets following formula respectively:
JQT≤SLA
JQT；
JET≤SLA
JET；
JTT≤SLA
JTT。
3. as claimed in claim 1 based on data center's resources of virtual machine dispatching method of segmentation service-level agreement, it is characterized in that, described resources of virtual machine comprises cpu, internal memory and bandwidth, to the resources of virtual machine had, action comprises increase resource, resource is constant and reduce resource, and three kinds of actions are expressed as 1,0 and-1.
4., as claimed in claim 1 based on data center's resources of virtual machine dispatching method of segmentation service-level agreement, it is characterized in that, to distribute and to reclaim as intensified learning motion space, Reward Program is described below immediately:
(1) if the UUTC of current work is greater than average UUTC, and this operation meets service-level agreement and QoS constraint, then return is 1;
(2) if current work does not meet service-level agreement and QoS constraint, then return is for-1;
(3) other situations, return is 0.
5., as claimed in claim 1 based on data center's resources of virtual machine dispatching method of segmentation service-level agreement, it is characterized in that, the rule of formulation is:
(1) if operation certain stage in the process of implementation, violate the constraint of segmentation service-level agreement, then this operation is in follow-up implementation, increases the resources of virtual machine distributed;
(2) if the UUTC of this operation is less than average UUTC, then this operation is in follow-up implementation, subtracts under absorbed resources of virtual machine;
Wherein, cpu resource increases at every turn or reduces 1, and internal memory increases at every turn or reduces 256M, and bandwidth increases at every turn or reduces 256kbps.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510733449.3A CN105260230B (en) | 2015-10-30 | 2015-10-30 | Data center's resources of virtual machine dispatching method based on segmentation service-level agreement |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510733449.3A CN105260230B (en) | 2015-10-30 | 2015-10-30 | Data center's resources of virtual machine dispatching method based on segmentation service-level agreement |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105260230A true CN105260230A (en) | 2016-01-20 |
| CN105260230B CN105260230B (en) | 2018-06-26 |
Family
ID=55099933
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510733449.3A Active CN105260230B (en) | 2015-10-30 | 2015-10-30 | Data center's resources of virtual machine dispatching method based on segmentation service-level agreement |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105260230B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107070965A (en) * | 2016-12-22 | 2017-08-18 | 广东石油化工学院 | A kind of Multi-workflow resource provision method virtualized under container resource |
| CN108476084A (en) * | 2016-12-02 | 2018-08-31 | 华为技术有限公司 | The method and apparatus on adjustment state space boundary in Q study |
| CN109587217A (en) * | 2018-11-06 | 2019-04-05 | 华中科技大学 | A kind of intelligent monitoring of large scale distributed system and management method and system |
| CN109725988A (en) * | 2017-10-30 | 2019-05-07 | 北京京东尚科信息技术有限公司 | A kind of method for scheduling task and device |
| CN109901929A (en) * | 2019-03-04 | 2019-06-18 | 云南大学 | Cloud computing task share fair allocat method under server level constraint |
| CN110362411A (en) * | 2019-07-25 | 2019-10-22 | 哈尔滨工业大学 | A kind of cpu resource dispatching method based on Xen system |
| CN110413396A (en) * | 2019-07-30 | 2019-11-05 | 广东工业大学 | A kind of resource regulating method, device, equipment and readable storage medium storing program for executing |
| CN110737529A (en) * | 2019-09-05 | 2020-01-31 | 北京理工大学 | cluster scheduling adaptive configuration method for short-time multiple variable-size data jobs |
| CN113610372A (en) * | 2021-07-27 | 2021-11-05 | 远景智能国际私人投资有限公司 | Method, device, terminal and readable storage medium for determining service level agreement |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050262504A1 (en) * | 2004-05-21 | 2005-11-24 | Esfahany Kouros H | Method and apparatus for dynamic CPU resource management |
| CN102611622A (en) * | 2012-02-28 | 2012-07-25 | 清华大学 | Dispatching method for working load of elastic cloud computing platform |
| CN103559089A (en) * | 2013-10-30 | 2014-02-05 | 南京邮电大学 | Method for implementing forecasting on requirement of virtual machine on basis of service level agreement constraints |
| CN103607459A (en) * | 2013-11-21 | 2014-02-26 | 东北大学 | Dynamic resource monitoring and scheduling method of cloud computing platform IaaS layer |
| CN104135535A (en) * | 2014-08-14 | 2014-11-05 | 苏州大学 | Tenant adjusting method and system oriented to cloud computing |
-
2015
- 2015-10-30 CN CN201510733449.3A patent/CN105260230B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050262504A1 (en) * | 2004-05-21 | 2005-11-24 | Esfahany Kouros H | Method and apparatus for dynamic CPU resource management |
| CN102611622A (en) * | 2012-02-28 | 2012-07-25 | 清华大学 | Dispatching method for working load of elastic cloud computing platform |
| CN103559089A (en) * | 2013-10-30 | 2014-02-05 | 南京邮电大学 | Method for implementing forecasting on requirement of virtual machine on basis of service level agreement constraints |
| CN103607459A (en) * | 2013-11-21 | 2014-02-26 | 东北大学 | Dynamic resource monitoring and scheduling method of cloud computing platform IaaS layer |
| CN104135535A (en) * | 2014-08-14 | 2014-11-05 | 苏州大学 | Tenant adjusting method and system oriented to cloud computing |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108476084B (en) * | 2016-12-02 | 2020-05-08 | 华为技术有限公司 | Method and device for adjusting state space boundary in Q learning |
| CN108476084A (en) * | 2016-12-02 | 2018-08-31 | 华为技术有限公司 | The method and apparatus on adjustment state space boundary in Q study |
| CN107070965B (en) * | 2016-12-22 | 2021-02-19 | 广东石油化工学院 | Multi-workflow resource supply method under virtualized container resource |
| CN107070965A (en) * | 2016-12-22 | 2017-08-18 | 广东石油化工学院 | A kind of Multi-workflow resource provision method virtualized under container resource |
| CN109725988A (en) * | 2017-10-30 | 2019-05-07 | 北京京东尚科信息技术有限公司 | A kind of method for scheduling task and device |
| CN109587217A (en) * | 2018-11-06 | 2019-04-05 | 华中科技大学 | A kind of intelligent monitoring of large scale distributed system and management method and system |
| CN109901929A (en) * | 2019-03-04 | 2019-06-18 | 云南大学 | Cloud computing task share fair allocat method under server level constraint |
| CN109901929B (en) * | 2019-03-04 | 2023-04-18 | 云南大学 | Cloud computing task share fair distribution method under server level constraint |
| CN110362411A (en) * | 2019-07-25 | 2019-10-22 | 哈尔滨工业大学 | A kind of cpu resource dispatching method based on Xen system |
| CN110413396A (en) * | 2019-07-30 | 2019-11-05 | 广东工业大学 | A kind of resource regulating method, device, equipment and readable storage medium storing program for executing |
| CN110413396B (en) * | 2019-07-30 | 2022-02-15 | 广东工业大学 | Resource scheduling method, device and equipment and readable storage medium |
| CN110737529A (en) * | 2019-09-05 | 2020-01-31 | 北京理工大学 | cluster scheduling adaptive configuration method for short-time multiple variable-size data jobs |
| CN110737529B (en) * | 2019-09-05 | 2022-02-08 | 北京理工大学 | Short-time multi-variable-size data job cluster scheduling adaptive configuration method |
| CN113610372A (en) * | 2021-07-27 | 2021-11-05 | 远景智能国际私人投资有限公司 | Method, device, terminal and readable storage medium for determining service level agreement |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105260230B (en) | 2018-06-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105260230A (en) | Resource scheduling method for data center virtual machine based on segmented service level agreement | |
| Priya et al. | Resource scheduling algorithm with load balancing for cloud service provisioning | |
| Zuo et al. | A multi-objective optimization scheduling method based on the ant colony algorithm in cloud computing | |
| CN109582448B (en) | Criticality and timeliness oriented edge calculation task scheduling method | |
| Sun et al. | PACO: A period ACO based scheduling algorithm in cloud computing | |
| CN108182115A (en) | A kind of virtual machine load-balancing method under cloud environment | |
| Xu et al. | Migration cost and energy-aware virtual machine consolidation under cloud environments considering remaining runtime | |
| CN105446816B (en) | A kind of energy optimization dispatching method towards heterogeneous platform | |
| CN104331321A (en) | Cloud computing task scheduling method based on tabu search and load balancing | |
| CN110308967A (en) | A kind of workflow cost based on mixed cloud-delay optimization method for allocating tasks | |
| Liu et al. | A survey on virtual machine scheduling in cloud computing | |
| CN107370799B (en) | A kind of online computation migration method of multi-user mixing high energy efficiency in mobile cloud environment | |
| CN104536804A (en) | Virtual resource dispatching system for related task requests and dispatching and distributing method for related task requests | |
| CN107168770A (en) | A kind of cloud data center workflow schedule of low energy consumption and resource provision method | |
| Nan et al. | A New Task Scheduling Scheme Based on Genetic Algorithm for Edge Computing. | |
| CN106802822A (en) | A kind of cloud data center cognitive resources dispatching method based on moth algorithm | |
| Khatua et al. | Prediction-based instant resource provisioning for cloud applications | |
| Pandya et al. | Dynamic resource allocation techniques in cloud computing | |
| Qin et al. | Joint energy optimization on the server and network sides for geo-distributed data centers | |
| CN117493020A (en) | Method for realizing computing resource scheduling of data grid | |
| CN112306642A (en) | Workflow scheduling method based on stable matching game theory | |
| CN105138391B (en) | The multitasking virtual machine distribution method of cloud system justice is distributed towards wide area | |
| Cao et al. | Online cost-rejection rate scheduling for resource requests in hybrid clouds | |
| Li et al. | Resource scheduling approach for multimedia cloud content management | |
| Manikandan et al. | Cost Effective Optimal Task Scheduling Model in Hybrid Cloud Environment. |
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 |