Research On Task Scheduling Policy Driven By Economy In Grid Environment

In the grid environment, task scheduling is the key problem to be solved,which also is the foundation of grid application. Meanwhile, scheduling policy is the most important aspect in task scheduling. A good policy not only can reduce task execution time and cost and improve the quality of service, but also can effectively utilize the massive heterogeneous resources on the grid and attract more resources owners to devote their resources to grid. Thereinto, task scheduling policy based on the economy are increasingly being recognized as an effective scheduling policy in grid environment.This paper mainly focuses on the task scheduling policy driven by the economy, elaborates the Nimrod/G grid resource management system and a set of scheduling policy with DBC (Deadline and Budget Constrained) adopted in the framework. To overcome shortcomings of CO (Cost Optimization) scheduling policy,low task completion rate and unbalance resource load,FCO(Fuzzy Cost Optimization)scheduling policy is brought forward based on DBC.The FCO scheduling policy firstly classifies the resources according to the relevance of using cost and then chooses the most reasonable resources for the task according to the load and computation ability of resource, which can complete more tasks with the limitation of expected execution time and cost, and with fewer price and load.In order to examine the validity of the FCO scheduling policy, several simulated experiments were carried on with the grid simulator GridSim. The simulation results show that the FCO scheduling strategy proposed in this paper is reasonable; this algorithm can provide better flexibility for users, satisfy actual need of them. And meanwhile, this policy completes more tasks with the limited time and budget and effectively solves the problem of unbalanced load.