Research On Batch-mode Job Scheduling Algorithms In Environment Of Grid

Grid computing, emerging as a new paradigm for next-generation computing, enables the aggregation of geographically distributed heterogeneous resources for solving large-scale problems in science, engineering, and commerce. So the task scheduling in Grid has become a quite important problem. And the research of schedule algorithms, which directly related to the speed, quality and other factors of the Grid, plays a decisive role in the research of computing Grid.Job scheduling mode include online mode and batch mode in time aspect analyzing.The former is the job is send to resource immediately after the job is submitted by user,this schedule mode have a good real time,but the scheduling effect is bad.The latter is jobs are scheduled after scheduling cycle or arriving some jobs,this mode a worse real time,but this schedule algorithms take jobs muster and resource muster into account,and can get a better scheduling effect.This paper adopt batch mode scheduling algorithms.In grid computing,we always decompose some large jobs into a lot of small jobs,whose run time is very short,and schedule job is required time,so each small job that is scheduled one time may reduce efficiency.Now batch-mode scheduling algorithm can primely solve this problem.Using batch-mode, job that has just arrived is not mapped to resource immediately,but waiting other jobs to group into a job set,after the map accident is coming,the resource schedule module map all jobs of this set to resource together.But batch-mode don't take the waiting time of arrived jobs into accout.Though it can save the schedule time of job, if all jobs waiting too long in a job set,efficiency can go down instead.Due to the contradiction of schedule time and wait time,this paper focuses on some researched and improving on this algorithms:firstly, discusses the significance of the grid job schedule algorithms,introduces the interrelated research and status of grid job scheduling algorithms and jobs scheduling algorithms simulator. Secondly,we analyse two existed batch-mode scheduling algorithms and its deficiency.Due to the change of scheduling time,we adopt the method of dynamic setting batch jobs size, and bring forward the module of the impoving algorithms.Finally, base on the impoving algorithms,we modify the code of the GridSim simulator,and simulate this algorithms.Through comparing some group experiment,we analyse and contrast from many points of view,and prove that improving algorithms is moreeffective and more stability, more adaptive to the dynamic environment of grid.21pitures, 3 tables, 44 referrences.