In existing web services-based workflow, data exchanging across the web services is centralized, the workflow engine intermediates at each step of the application sequence. However, many grid applications, especially ...In existing web services-based workflow, data exchanging across the web services is centralized, the workflow engine intermediates at each step of the application sequence. However, many grid applications, especially data intensive scientific applications, require exchanging large amount of data across the grid services. Having a central workflow engine relay the data between the services would resu＇lts in a bottleneck in these cases. This paper proposes a data exchange model for individual grid workflow and multiworkflows composition respectively. The model enables direct communication for large amounts of data between two grid services. To enable data to exchange among multiple workflows, the bridge data service is used.展开更多
In order to effectively control the random tasks submitted and executed in grid workflow,a grid workflow model based on hybrid petri-net is presented. This model is composed of random petri-net,colored petri-net and g...In order to effectively control the random tasks submitted and executed in grid workflow,a grid workflow model based on hybrid petri-net is presented. This model is composed of random petri-net,colored petri-net and general petri-net. Therein random petri-net declares the relationship between the number of grid users' random tasks and the size of service window and computes the server intensity of grid system. Colored petri-net sets different color for places with grid services and provides the valid interfaces for grid resource allocation and task scheduling. The experiment indicated that the model presented in this letter could compute the valve between the number of users' random tasks and the size of grid service window in grid workflow management system.展开更多
Accurate performance prediction of Grid workflow activities can help Grid schedulers map activitiesto appropriate Grid sites.This paper describes an approach based on features-ranked RBF neural networkto predict the p...Accurate performance prediction of Grid workflow activities can help Grid schedulers map activitiesto appropriate Grid sites.This paper describes an approach based on features-ranked RBF neural networkto predict the performance of Grid workflow activities.Experimental results for two kinds of real worldGrid workflow activities are presented to show effectiveness of our approach.展开更多
With the emergence of grid computing, new challenges have arisen in workflow tasks scheduling. The goal of grid-workflow task scheduling is to achieve high system throughput and to match the application needs with the...With the emergence of grid computing, new challenges have arisen in workflow tasks scheduling. The goal of grid-workflow task scheduling is to achieve high system throughput and to match the application needs with the available computing resources. This matching of resources in a non-deterministically share hererogeneous environment leads to concerns on quality of service (QoS). Grid concept is presented in this paper, coupled with the QoS requirement of workflow task and an improved algorithm-ILGSS algorithm, has been brought out. The complexity of the improved scheduling algorithm has been analyzed. The experiment results show that the improved algorithm can lead to significant performance gain in various applications. An important research domain-adaptive workflow transaction in grid computing environment, has been explored and a new solution for the scheduling of distribute workflow has been bring forward in grid environment. Key words grid - workflow - OGSA - adaptive - QoS CLC number TP 393 Foundation item: Supported by the National Natural Science Foundation of China(60073050)Biography: ZHANG Yun-feng (1981-), male, Master candidate, research direction: software engineering, grid computiong.展开更多
The burgeoning grid technology provides a practical possibility to share Numerical Control （NC） equipment under a Web environment via many-w-many mode which is faced with the difficulty of a driving force for grid w...The burgeoning grid technology provides a practical possibility to share Numerical Control （NC） equipment under a Web environment via many-w-many mode which is faced with the difficulty of a driving force for grid workflow. This paper proposes the novel methodology of coupling constraints to overcome this difficulty. First, the complex constraints and the structure attributes of NC tasks are detailed in the NC equipment grid that illustrates the possibility of forming a grid workflow force by coupling constraints. Second, the principle of driving force by coupling constraints in an NC equipment grid and the relative algorithm are proposed. Finally, the case study of comparative experiment is detailed to demonstrate the efficiency of this novel tactic.展开更多
针对成本约束有向无环图DAG(directed acyclic graph)表示的网格工作流完工时间最小化问题,提出两个基于优先级规则的迭代启发算法.算法利用并行活动特征定义正向分层和逆向分层两个概念,将其分别引入最大收益规则MP(maximum profit),...针对成本约束有向无环图DAG(directed acyclic graph)表示的网格工作流完工时间最小化问题,提出两个基于优先级规则的迭代启发算法.算法利用并行活动特征定义正向分层和逆向分层两个概念,将其分别引入最大收益规则MP(maximum profit),得到正分层最大收益规则MPTL(maximum profit with top level)和逆分层最大收益规则MPBL(maximum profit with bottom level).两规则每次迭代尽量以完工时间的最小增加换取总费用的最大降低,逐步将分层初始解构造为满足成本约束的可行解.模拟结果表明,两规则在获得较少迭代次数和运行时间的同时,能显著改进MP规则的平均性能,且MPBL优于MPTL.展开更多
基金Supported by the National Natural Science Foun-dation of China(60373072)
文摘In existing web services-based workflow, data exchanging across the web services is centralized, the workflow engine intermediates at each step of the application sequence. However, many grid applications, especially data intensive scientific applications, require exchanging large amount of data across the grid services. Having a central workflow engine relay the data between the services would resu＇lts in a bottleneck in these cases. This paper proposes a data exchange model for individual grid workflow and multiworkflows composition respectively. The model enables direct communication for large amounts of data between two grid services. To enable data to exchange among multiple workflows, the bridge data service is used.
基金the National Natural Science Foundation of China (No.60573141, 70271050)the Natural Science Foundation of Jiangsu Province (No.BK2005146)+3 种基金the High Technology Research Programme of Jiangsu Prov-ince (No.BG2005037, BG2005038, BG2006001)the High Technology Research Programme of Nanjing (No. 2006RZ105)the Foundation of National Laboratory for Modern Communications (No.9140C1101010603)the Key Laboratory of Information Technology Processing of Jiangsu Province (No.kjs05001, kjs0606).
文摘In order to effectively control the random tasks submitted and executed in grid workflow,a grid workflow model based on hybrid petri-net is presented. This model is composed of random petri-net,colored petri-net and general petri-net. Therein random petri-net declares the relationship between the number of grid users' random tasks and the size of service window and computes the server intensity of grid system. Colored petri-net sets different color for places with grid services and provides the valid interfaces for grid resource allocation and task scheduling. The experiment indicated that the model presented in this letter could compute the valve between the number of users' random tasks and the size of grid service window in grid workflow management system.
基金Supported by the European Union through the IST-034601 edutain@grid project
文摘Accurate performance prediction of Grid workflow activities can help Grid schedulers map activitiesto appropriate Grid sites.This paper describes an approach based on features-ranked RBF neural networkto predict the performance of Grid workflow activities.Experimental results for two kinds of real worldGrid workflow activities are presented to show effectiveness of our approach.
文摘With the emergence of grid computing, new challenges have arisen in workflow tasks scheduling. The goal of grid-workflow task scheduling is to achieve high system throughput and to match the application needs with the available computing resources. This matching of resources in a non-deterministically share hererogeneous environment leads to concerns on quality of service (QoS). Grid concept is presented in this paper, coupled with the QoS requirement of workflow task and an improved algorithm-ILGSS algorithm, has been brought out. The complexity of the improved scheduling algorithm has been analyzed. The experiment results show that the improved algorithm can lead to significant performance gain in various applications. An important research domain-adaptive workflow transaction in grid computing environment, has been explored and a new solution for the scheduling of distribute workflow has been bring forward in grid environment. Key words grid - workflow - OGSA - adaptive - QoS CLC number TP 393 Foundation item: Supported by the National Natural Science Foundation of China(60073050)Biography: ZHANG Yun-feng (1981-), male, Master candidate, research direction: software engineering, grid computiong.
基金supported by NSFC under Grant No.50705072Hubei Province Natural Science Fund for Young Scientist under Grant No.2005ABB023Wuhan City Dawn Planunder Grant No.20055003059
文摘The burgeoning grid technology provides a practical possibility to share Numerical Control （NC） equipment under a Web environment via many-w-many mode which is faced with the difficulty of a driving force for grid workflow. This paper proposes the novel methodology of coupling constraints to overcome this difficulty. First, the complex constraints and the structure attributes of NC tasks are detailed in the NC equipment grid that illustrates the possibility of forming a grid workflow force by coupling constraints. Second, the principle of driving force by coupling constraints in an NC equipment grid and the relative algorithm are proposed. Finally, the case study of comparative experiment is detailed to demonstrate the efficiency of this novel tactic.
文摘针对成本约束有向无环图DAG(directed acyclic graph)表示的网格工作流完工时间最小化问题,提出两个基于优先级规则的迭代启发算法.算法利用并行活动特征定义正向分层和逆向分层两个概念,将其分别引入最大收益规则MP(maximum profit),得到正分层最大收益规则MPTL(maximum profit with top level)和逆分层最大收益规则MPBL(maximum profit with bottom level).两规则每次迭代尽量以完工时间的最小增加换取总费用的最大降低,逐步将分层初始解构造为满足成本约束的可行解.模拟结果表明,两规则在获得较少迭代次数和运行时间的同时,能显著改进MP规则的平均性能,且MPBL优于MPTL.
