CN101127700A - Grid resource scheduling method based on network bandwidth and user charge - Google Patents
Grid resource scheduling method based on network bandwidth and user charge Download PDFInfo
- Publication number
- CN101127700A CN101127700A CNA200710052966XA CN200710052966A CN101127700A CN 101127700 A CN101127700 A CN 101127700A CN A200710052966X A CNA200710052966X A CN A200710052966XA CN 200710052966 A CN200710052966 A CN 200710052966A CN 101127700 A CN101127700 A CN 101127700A
- Authority
- CN
- China
- Prior art keywords
- resource
- task
- bandwidth
- resources
- list
- 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
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The utility model relates to a scheduling method of the grid resource based on the network bandwidth and at the expense of users. The method comprises the following steps: after identifying a resource, the resource agent adds the resource and the task to the resource list and the task list respectively; then arranged the resources in an ascending order according to the prices, with a view to assign the tasks to the less expensive resources as possible during the allocation of the resources to lower the costs for the users; when allocating the resources, the agent should take into consideration the demands of the network bandwidth; if the bandwidth provided by the resources can meet the requirements of the tasks against the bandwidth, the agent should allocate the resource to the corresponding task; a mapping relationship should be established between the resources and tasks to accomplish all the tasks at the lowest costs as possible; the steps can be repeated until all the tasks are properly allotted. By striving to complete the tasks submitted by users at the desirable costs, the utility model can also perfectly satisfy the requirements of network bandwidth.
Description
Technical field
The invention belongs to a kind of grid resource scheduling method, the grid resource scheduling method under particularly a kind of bandwidth Network Based and the customer charge.
Background technology
Along with the fast development of Internet and improving constantly of computer technology, and problem solving more and more develops towards high-performance, extensive, diversity, direction such as multi-functional.The various network resources that how will be distributed in diverse geographic location puts together combined calculation, solve the same problem to become and become more and more important, and produced " grid computing " thus.Grid is exactly integrated calculating and resource environment, in other words conj.or perhaps a computational resource pond.Lan Foseter points out that grid computing has in the Virtual Organization of a plurality of departments or group in dynamic change, carries out collaborative flexibly resource-sharing of safety and problem solving.Gridding resource is meant the general name of all entities that can be utilized in the grid.In the grid environment, resource is distributed on the diverse geographic location, is had and is operated by different tissues or individual, and the Resource Owner has absolute power to resource, and grid must satisfy local management strategy to its management.Resource owner can also determine resource whether to add grid at any time or withdraw from grid, and the load of gridding resource also is dynamic, and grid must reflect the requirement that resource dynamic changes.Simultaneously, the type of gridding resource is very extensive, One's name is legion.Characteristics at this autonomy of gridding resource, isomerism, dynamic, complexity, adopt which kind of management mode that gridding resource is managed and dispatches to become a key issue in the grid computing, many scholars and scientific research institution have also dropped into great amount of manpower and material resources to this problem, have carried out deep research.
At present, grid resource and scheduling model mainly concentrate on three kinds of patterns: centralized, distributing, stagewise.When management complex environment like this, traditional method for managing resource of attempting the optimization global system is worthless, because the centralized strategy that conventional method is used needs state information and common structure management strategy completely.For the computing grid of setting up a success is impossible.Therefore stagewise and distributed management method are suitable to the management and the running of gridding resource.In these methods, it is to utilize economic model to manage supply-demand relationship with projected resources that certain methods is arranged.Gridding resource is acted on behalf of GRB (Grid Resource Broker) and is coordinated mutual between grid resource owner GRP (playing the part of traditional mode of production person Grid Resource Producer) and the grid user GRC (represent consumer Grid Resource Consumer), and resource is united by the middleware that moves low level and joined in the grid.The core middleware support resource access registrar that on gridding resource owner resource, moves, and only allow through the user capture of mandate they.Grid application is created and is carried out in the core middleware support that moves on the resource user.Resource Broker is then finished resource discovering, selection, and set, data and program transmission, and work such as initialization task execution and collection on remote resource, also responsible monitor task executive process and managing gridding environmental change and the resource that may occur are invalid etc.
Superiority based on the economic model scheduling of resource is the scalable imbalance between supply and demand, because making the process of resource dispatching strategy decision is distributed on one's body all Resource consumers and resource provider, therefore make scheduling by being that the center has turned to customer-centric with the system, the user can make their own decision and obtain best performance with the cost of minimum.Also can help the developer to develop scheduling strategy in addition, and then set up a height open-ended system.This model be suitable for most computing grid dynamically and isomery hold and levy, but at first must seek a resource provisioning and require function, embodying Resource Owner and Resource consumers both sides' interests.But conflict often appears in this requirement in both parties.Therefore many scholars are placed on how to pass through to add various QoS indexs to research emphasis, balance Resource Owner and resource user two sides' interests, many algorithms have also been worked out, wherein, worked out following classical dispatching method with the research team headed by the famous grid expert Rajkumar Buyya of Univ Melbourne Australia: the time optimal under the cost optimal algorithm under the time-constrain, the cost constraint, the time cost optimum under the time cost constraint.These three kinds of algorithms all be from the angle of application layer consider the user qos requirement, the deadline that resource user's cost and task are finished is taken all factors into consideration, reach the scheduling purpose of an optimum.Yet grid is the architecture of a multilayer, not only comprises application layer, also has techonosphere, articulamentum, and resource layer, convergence-level, application layer are the top of mesh architecture, directly towards grid user.The resource of grid system is based upon on the high performance network with scheduling, and the qos requirement of network is directly restricting the application of grid, and in a sense, the qos requirement of network becomes the bottleneck of grid application.In diverse network QoS, what at first should consider is the bandwidth of network.Therefore, only consider that the qos requirement of application layer can not solve the problem that exists in the grid resource scheduling at all, also do not reach the effect that improves whole grid throughput.
Summary of the invention
The purpose of this invention is to provide a kind of thrifty task of finishing the user of trying one's best, consider the network bandwidth, improve the bandwidth Network Based of grid system throughput and the grid resource scheduling method under the customer charge, to overcome above-mentioned deficiency.
To achieve these goals, the method applied in the present invention is:
First step: Resource Broker searches out suitable gridding resource;
Second step: resource and task are joined respectively in the Resources list and the task list;
Third step: resource is arranged according to the price ascending order, and purpose is when Resources allocation, as far as possible with Task Distribution on the lower resource of price, to save customer charge;
The 4th step: when Resources allocation, consider the demand of the network bandwidth,, then give this task with resource allocation if the bandwidth that resource can provide can satisfy the demand of task to bandwidth;
The 5th step: with resource and task creation mapping relations, the task that makes can both be finished, and the energy guarantee fee is with thrifty;
The 6th step: repeat above step, all distribute up to all tasks.
Adopt the key problem that economic model manages and dispatches gridding resource becomes grid research, many scholars and research institution have also proposed some solutions.But these methods mainly go to consider from user's angle, as user's expense, time deadline etc., however grid computing be based upon on the express network basis, do not have the assurance of the network bandwidth, just can't realize the scheduling of grid computing and gridding resource.Therefore, when research grid resource and scheduling, must consider the constraint of the network bandwidth.The present invention proposes the grid resource scheduling method under bandwidth Network Based and the customer charge, not only consider and on expense, finish the task that the user submits to as far as possible thriftily, guaranteed the demand of the network bandwidth simultaneously.
The present invention compares with the conventional mesh scheduling of resource, the advantage of performance is: (1) has taken into full account the demand of grid multi-tier systematic structure, comprise the customer charge of application layer and the network bandwidth of techonosphere, and the traditional scheduler method is only considered the user's request of application layer; (2) grid resource scheduling method under a kind of bandwidth Network Based and the customer charge has been proposed, considered that grid computing is to be based upon on the basis of express network, in the task that the user submitted to, many have specific (special) requirements to the network bandwidth, therefore, the present invention has guaranteed the bandwidth demand of task.And conventional method has also only been considered user's demand, bandwidth is not considered, therefore many tasks with bandwidth demand are actually inexecutable.So the present invention has very strong realistic meaning; (3) the present invention proposes a kind of non-resource regulating method of mapping one by one, resource and task can be the relations of one-to-many, promptly for low-cost resource, can give it with more Task Distribution, can save user's expense so as far as possible.
Description of drawings
Fig. 1 is based on calculating economic grid resource scheduling device illustraton of model.
Fig. 2 is a flow chart of the present invention.
Embodiment
The present invention is described in further detail below in conjunction with accompanying drawing.
In the grid environment, resource is meant and is distributed on the diverse geographic location to have autonomy, isomerism, dynamic, complexity, can satisfy the software and hardware resources of the different demands of grid user.Resource can be with one group of vector representation R (R
1, R
2, R
3, R
4... R
m), it is embodying Resource Owner's requirement, and promptly price and these two factors of bandwidth that can provide all are provided each resource.
In the grid environment, the Resource Owner sells the resource of oneself, obtains economic interests, and the resource user buys resource and finishes the task of oneself.In economic model, the will of both parties is to get off to realize that scheduling process is as follows in the scheduling of scheduler (as Fig. 1):
1, the user sets up a task (job), and task comprises the description to user's request, and all tasks are formed a series of gridlet tabulations, and user's demand can be passed to Resource Scheduler by application programming interfaces;
2, the discovery of resource and transaction modules and GIS (Grid Information Servers) module interacts and determines the information of resource, set up the configuration and the visit cost (Cost) of resource then with the resource interaction, set up succedaneum's the Resources list like this, operation in the gridlet tabulation is carried out in the Resources list, by test and deduction method, the resource performance data are used as prediction;
3, dispatching management module is selected adequate resources according to user's demand (customer charge, bandwidth constraint) for gridlet according to resource scheduling algorithm;
4, for each resource, the gridlet distribution module selects the quantity of gridlet to carry out to avoid the operation for the sole user according to usage policy, and the load of resource is overweight;
5, the gridlet distribution module is submitted to selected resource with it;
6, after the grdlet processing finishes, resource is instead gone back gridlet and is accepted module to gridlet, accepts module testing and upgrades operation time parameters;
7, repeat 3-6, only finish, return result then and give the user to all gridlet are processed.
And the present invention has considered two QoS constraints of the network bandwidth of the customer charge and the techonosphere of application layer, under these two QoS constraints, finish user's task the most thriftily, its basic thought is exactly that resource is arranged by the price ascending order, assigns the task to resource then under the user bandwidth constraint.Its concrete grammar is:
(1) resource is joined list of available resources, and arrange according to the price ascending order;
(2) task is added task list successively;
(3) for each resource that sorts, from task list, take out a task, if the bandwidth of this mission requirements is not more than the bandwidth of lowest price resource in the Resources list, give this resource, and task is deleted from task list this Task Distribution;
(4) if the bandwidth of its requirement greater than the bandwidth of lowest price resource in the Resources list, attempt so to give next resource this Task Distribution, if satisfy bandwidth requirement, be allocated successfully, otherwise seek next resource successively, up to finding adequate resources, and this task is deleted from task list;
(5) repeat above steps, intact up to all Task Distribution.
User task is embodying the requirement description of grid user to gridding resource, with one group of vector representation task T (T
1, T
2, T
3, T
4... T
m), comprising user's various qos requirements, what the present invention will consider is user's expense (cost) and the restriction (bandwidth) of user to finishing this required by task bandwidth.Simultaneously, each task has again by several subtasks and constitutes, and the subtask represents with gridlet, i.e. T
i(gridlet
1, gridlet
2, gridlet
3Gridlet
n), 1≤i≤m, n 〉=1.
A, the resource in the Resources list is arranged according to the price ascending order, guaranteed that Task Distribution is to assign the task to low-cost resource as far as possible, saves customer charge.
The Resources list E={R[1], R[2] ... R[n] }
R[1].cost≤R[2].cost…≤R[n].cost…………………………①
B, when Task Distribution, guarantee to finish the bandwidth demand of this task, promptly the resource bandwidth that can provide can not be less than the bandwidth of finishing this required by task
G[i].bandwidth≤R[j].bandwidth……………………………………②
By formula 1. as can be known, when giving task, preferentially low-cost resource is distributed to task with resource allocation, therefore can greatly save the user expense.
By formula 2. as can be known, the bandwidth of having only resource to provide is not less than when finishing the work required bandwidth, could give task with this resource allocation, therefore guarantee the demand of task to the network bandwidth.
By above analysis as can be known, the task one that the user submitted to satisfies its demand to bandwidth surely, and cost saving.
The content that is not described in detail in the specification of the present invention belongs to this area professional and technical personnel's known prior art.
Claims (2)
1. the grid resource scheduling method under bandwidth Network Based and the customer charge, its method is:
First step: Resource Broker searches out suitable gridding resource;
Second step: resource and task are joined respectively in the Resources list and the task list;
Third step: resource is arranged according to the price ascending order, and purpose is when Resources allocation, as far as possible with Task Distribution on the lower resource of price, to save customer charge;
The 4th step: when Resources allocation, consider the demand of the network bandwidth,, then give this task with resource allocation if the bandwidth that resource can provide can satisfy the demand of task to bandwidth;
The 5th step: with resource and task creation mapping relations, the task that makes can both be finished, and the energy guarantee fee is with thrifty;
The 6th step: repeat above step, all distribute up to all tasks.
2. the grid resource scheduling method under bandwidth Network Based as claimed in claim 1 and the customer charge is characterized in that: its concrete grammar is:
The first step: resource is joined list of available resources, and arrange according to the price ascending order;
Second step: task is added task list successively;
The 3rd step: for each resource that sorts, from task list, take out a task, if the bandwidth of this mission requirements is not more than the bandwidth of lowest price resource in the Resources list, give this resource, and task is deleted from task list with this Task Distribution;
The 4th step: if the bandwidth of its requirement is greater than the bandwidth of lowest price resource in the Resources list, attempt so to give next resource this Task Distribution, if satisfy bandwidth requirement, be allocated successfully, otherwise seek next resource successively, up to finding adequate resources, and this task is deleted from task list;
The 5th step: repeat above steps, intact up to all Task Distribution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA200710052966XA CN101127700A (en) | 2007-08-16 | 2007-08-16 | Grid resource scheduling method based on network bandwidth and user charge |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA200710052966XA CN101127700A (en) | 2007-08-16 | 2007-08-16 | Grid resource scheduling method based on network bandwidth and user charge |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101127700A true CN101127700A (en) | 2008-02-20 |
Family
ID=39095627
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA200710052966XA Pending CN101127700A (en) | 2007-08-16 | 2007-08-16 | Grid resource scheduling method based on network bandwidth and user charge |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101127700A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101945434A (en) * | 2009-07-03 | 2011-01-12 | 华为技术有限公司 | Method, device and system for realizing load balance |
| CN102299842A (en) * | 2011-05-19 | 2011-12-28 | 江苏电力信息技术有限公司 | Grid resource co-allocation method capable of sensing production scheduling delay |
| TWI555423B (en) * | 2014-09-11 | 2016-10-21 | 國立交通大學 | Resource allocating method, base station, resource requesting method and user equipment |
| US10356802B2 (en) | 2017-12-20 | 2019-07-16 | Industrial Technology Research Institute | Base station and scheduling method of uplink resource unit |
| CN110198344A (en) * | 2019-05-05 | 2019-09-03 | 网宿科技股份有限公司 | A kind of resource regulating method and system |
| CN110460457A (en) * | 2018-05-08 | 2019-11-15 | 大唐移动通信设备有限公司 | A kind of data transmission method and device |
| CN111757354A (en) * | 2020-06-15 | 2020-10-09 | 武汉理工大学 | Multi-user slicing resource allocation method based on competitive game |
-
2007
- 2007-08-16 CN CNA200710052966XA patent/CN101127700A/en active Pending
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101945434A (en) * | 2009-07-03 | 2011-01-12 | 华为技术有限公司 | Method, device and system for realizing load balance |
| CN102299842A (en) * | 2011-05-19 | 2011-12-28 | 江苏电力信息技术有限公司 | Grid resource co-allocation method capable of sensing production scheduling delay |
| CN102299842B (en) * | 2011-05-19 | 2013-09-25 | 江苏电力信息技术有限公司 | Grid resource co-allocation method capable of sensing production scheduling delay |
| TWI555423B (en) * | 2014-09-11 | 2016-10-21 | 國立交通大學 | Resource allocating method, base station, resource requesting method and user equipment |
| US10356802B2 (en) | 2017-12-20 | 2019-07-16 | Industrial Technology Research Institute | Base station and scheduling method of uplink resource unit |
| CN110460457A (en) * | 2018-05-08 | 2019-11-15 | 大唐移动通信设备有限公司 | A kind of data transmission method and device |
| CN110460457B (en) * | 2018-05-08 | 2021-03-02 | 大唐移动通信设备有限公司 | Data transmission method and device |
| CN110198344A (en) * | 2019-05-05 | 2019-09-03 | 网宿科技股份有限公司 | A kind of resource regulating method and system |
| WO2020224022A1 (en) * | 2019-05-05 | 2020-11-12 | 网宿科技股份有限公司 | Resource scheduling method and system |
| EP3754944A4 (en) * | 2019-05-05 | 2020-12-23 | Wangsu Science & Technology Co., Ltd. | Resource scheduling method and system |
| US11153370B2 (en) | 2019-05-05 | 2021-10-19 | Wangsu Science & Technology Co., Ltd. | Resource scheduling method and system |
| CN111757354A (en) * | 2020-06-15 | 2020-10-09 | 武汉理工大学 | Multi-user slicing resource allocation method based on competitive game |
| US11716748B2 (en) | 2020-06-15 | 2023-08-01 | Wuhan University Of Technology | Multi-user slice resource allocation method based on competitive game |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xhafa et al. | Computational models and heuristic methods for Grid scheduling problems | |
| CN100570569C (en) | Operation cross-domain control method under the grid computing environment | |
| Wang et al. | Adaptive scheduling for parallel tasks with QoS satisfaction for hybrid cloud environments | |
| Ernemann et al. | Economic scheduling in grid computing | |
| CN105159762B (en) | Heuristic cloud computing method for scheduling task based on Greedy strategy | |
| CN101719931B (en) | Multi-intelligent body-based hierarchical cloud computing model construction method | |
| US9542223B2 (en) | Scheduling jobs in a cluster by constructing multiple subclusters based on entry and exit rules | |
| CN108108245B (en) | Hybrid scheduling method and system for cloud platform wide-node scientific workflow | |
| CN104123182B (en) | Based on the MapReduce task of client/server across data center scheduling system and method | |
| CN101127700A (en) | Grid resource scheduling method based on network bandwidth and user charge | |
| Hussain et al. | SLA-RALBA: cost-efficient and resource-aware load balancing algorithm for cloud computing | |
| CN100495984C (en) | Grid task dispatching method based on time and expense restriction | |
| CN102708003A (en) | Method for allocating resources under cloud platform | |
| CN108170530A (en) | A kind of Hadoop Load Balancing Task Scheduling methods based on mixing meta-heuristic algorithm | |
| Mehrotra et al. | Towards an autonomic performance management approach for a cloud broker environment using a decomposition–coordination based methodology | |
| Zhou et al. | Concurrent workflow budget-and deadline-constrained scheduling in heterogeneous distributed environments | |
| CN110321198A (en) | A kind of container cloud platform computing resource and Internet resources coordinated dispatching method and system | |
| Shojafar et al. | An efficient scheduling method for grid systems based on a hierarchical stochastic Petri net | |
| CN109117254A (en) | A kind of dispatching method and system of deep learning frame | |
| CN101222522B (en) | Gridding task scheduling method considering gridding task importance and time urgency | |
| Li et al. | Cost-efficient fault-tolerant workflow scheduling for deadline-constrained microservice-based applications in clouds | |
| Yang et al. | Multi-policy-aware MapReduce resource allocation and scheduling for smart computing cluster | |
| CN106095591A (en) | A kind of virtual machine two-stage optimizing management and running platform based on cloud computing | |
| Li et al. | Trust-based and qos demand clustering analysis customizable cloud workflow scheduling strategies | |
| CN102790715A (en) | Resource allocation system and method based on internet |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Open date: 20080220 |