Research On Key Techniques Of Dynamic Grid Computing Workflow

Grid computing workflow can bring effective support for the modern enterprise restructuring, business implementation of cross-regional and cross-industry. And the dynamic nature of grid computing environment, the richness and diversity of resources, have brought great opportunities for the development of workflow technology and applications.This paper will in detail analysis and research the dynamic problems of workflow technology when it applied to the grid computing environment. To the system structure of the grid dynamic workflow from modeling to implementation and the dynamic characteristics of the grid computing environment, study deeply into the core issues of dynamic workflow on the current grid technology and then put forward questions and answers. The problems need to be solved are as follows:(1) The logic and semantic elements of the Grid dynamic workflow model are not clear. Currently, most literature which are based on the dynamic supporting elements of the grid dynamic workflow model, is introduced some elements which will cause the redo and circulation by logic and semantic ambiguity, resulting in the overlapping of a logical level and semantic level, the confusion of the machine's automatic processing and manual participation in the decision-making process, and leading to the execution path error of the overall workflow in the process of executing the workflow.(2) Grid computing workflow model is insufficient to dynamic support and lack of workflow scheduling strategy for the full use of rich dynamic grid resources. The current Grid computing workflow system can not fully reflect the dynamic nature of grid computing environment, in the running, once the environment changes, it will be difficult for workflow model to expand effectively according to the dynamic information. With the deepening of grid applications, for some related applications, the execution of workflow model lacks optimization. The scheduling algorithm, which is corresponded to the model, can not fully play the advantages of grid computing workflow, and utilization of the grid resource is very low. Thus it caused the leisure and waste of the resources.(3)Lacking the control strategy for the data stream congestion of the grid computing workflow. In the instance study of the grid computing workflow, the research of control flow is more, while the literature for data flow is very little, especially the research for the transmission scheduling strategy of the intensive data streams. Because of data stream's various use domain in the network environment of heterogeneous difference, the quality of the service and use competition and other characteristics, the transmition of the data stream in the dynamic grid computing environment can not simply rely on the transmission control strategy of the underlying network. At the same time, the stability of the transmission and high transmission throughput is directly related to the implementation of the workflow efficiency and the smooth implementation. For data streaming rates, link bandwidth and receiver performance, if they do not match, it is likely to cause congestion, buffer overflow, etc. Seriouslly it may lead to the failure of the implementation of workflow tasks.(4)Lack of the planning and expectations of a global grid workflow execution path.By solving the above problems, we obtained the following results.(1)With the actual participation of the "grid dynamic workflow research in intelligent transportation system" project, this paper re-define the grid computing workflow, distinguish the the two core concepts of workflow logic and workflow semantic and analyze the the situation of the workflow dynamic adaptation in the environment of grid computing. It also fixed some inappropriate place in the related literature, studied the model which is suit for grid dynamic workflow to modeling , cited black box model to deal with the uncertainty of model definition phase and put forward the correct decompose rules which support dynamic workflow model. By focusing on researching the dynamic adaptability of the grid computing workflow in the intelligent transportation system ITS , it provided corresponding test algorithm.(2)On the base of studying Petri nets and WF-net, this paper puts forward a new method called DFP-net, which supports the dynamic changes of the grid workflow modeling. The experiment proved that it has the feasibility, and it can adapt better to the dynamic changes of grid computing environment than static modeling method such as WF-net. And then based on DFP-net model , this paper puts forward DFSA , a method which is used to calculating the dynamic adaptive scheduling of the workflow by Grid. Test results show that in the process of workflow execution, not only can it improve dynamic adaptive, but also make full use of the rich service resources of the grid and have better overall executive performance.(3)Introducing active queue management AQM technology to deal with dynamic scheduling of grid computing data streams in intensive work, which aims to avoid congestion. Based on the analysis of RED and FEM, the classical algorithm of AQM, this paper combines the fuzzy mathematics method and active queue management technology, and puts forward data flow congestion control scheduling algorithm of grid computing data streams in intensive work, and that is data flow scheduling algorithm DFSFA which is based on fuzzy control. This algorithm mainly inspects link and node properties, introducing the price mechanism to measure the situation of data circulated sending in the grid computing environment, and adjusts sending rate at the sender, At the same time it controls grid computing workflow node queue to lower the stable value level and effectively avoids congestion occurs, provide data stream“s stable transmission to adapt to the dynamic changes of the network and the grid service node. Simulation results show that DFSFA realized the goal that AQM can maintain smaller cohort, provide greater throughput and stable performance.(4)Analyzed the grid computing workflow execution's importance of the dynamic path planning, the paper puts forth that we can apply Ant colony optimization theory to the dynamic programming of workflow execution paths. Thus we designed the algorithm DGWFF, a dynamic grid computing workflow planning algorithm which is based on the ant colony optimization yuan of heuristic architecture. This algorithm puts emphasis on dynamic factors such as the grid services, node properties and network status, and it is bound to the workflow service. What's more, it can provide evidence and guidance for the selection of execution paths. Experimental results showed the path planned by DGWFF is effective, meanwhile it can provide a good basis and path for the dynamic workflow execution and bring better execution efficiency.