Research On CRP Method-based Grid Task Schedule Pattern

Grid Computing is a type of Distributed Computing, with the purpose to build auniform, integrated and virtualized computing resource, and to realize the resourcesharing, management and access across different organizations. Except numericalcalculation, Grid will implement cooperated working, workflow integration, datasharing and interoperation. Task scheduling is a fatal problem of Grid Computing, andits strategy and algorithm will decide the efficiency and feasibility of the tasksrunning on the Grid environment.Aiming at the task scheduling problem of Grid Computing, the thesis bringsforward a Grid task scheduling pattern basing on the Capacity Requirements Planning(CRP) method. Then the thesis designs and implements the key algorithms requiredand an extensible simulation environment.Firstly, the comparability between CRP method and Grid task scheduling isanalyzed and demonstrated, and the model of Grid task scheduling basing on CRPmethod is built. Also the definitions of Grid task and computing resource used in thismodel are accomplished.Next, three key algorithms of the task scheduling pattern are designed andimplemented, namely task decomposing algorithm, initial scheduling algorithm andautomatic adjusting algorithm. These algorithms will be used in the correspondingphases during the procedure of task scheduling. And the model of Grid task andcomputing resource is further detailed.To check and evaluate the task scheduling pattern and the algorithm, a Gridsimulation environment with the name Grid-CRP is implemented. The environment isfocus on task scheduling simulation, and takes extensibility and portability intoaccount. The analysis and evaluation on the simulation environment and algorithmresults are also pointed out.The summary and expectation are at the end of the thesis.