Research On Economic-based Job Scheduling Strategy For Grid Computing

Gird computing is the product of distributed and high-performance computing. Use grid technology, people could found a super system of numeration after connecting with server, memory system and internet, respectively. This super system provides high-quality service about computing and information. However, it is a complex undertaking to manage and schedule resource in the grid.In grid computing platforms, the geographic distribution of resources owned by different organizations with different usage polices, cost models and varying load and availability patterns is problematic. The producers (resource owners) and consumers (resource users) have different goals, objectives, and requirements so that some tradition resource manages and form of scheduling is not suitable here. To bring economy to grid system, and do research on how to utilize economy in grid job scheduling, is a research attraction in the grid job scheduling area.This paper do a deep study about grid job scheduling architecture based on economy model. Analysis of characteristic and scarcity by current architecture set out a hierarchical grid job scheduling architecture based on the concept of virtual market. This architecture to solve the problem of kinds of economical models mixed in one grid architecture by using hierarchical collection of trees. Author design a kind of price adjustor in the resource broker architecture. Use price to reflect the resource load dynamic which high or low. At the same time, the change of the resource price also dynamic influence load of resource. The use of this solution to balance the resource load and to solve the problem of resource price never changed in job scheduling process of tradition grid architecture. Author also analysis the scheduling influence of creditable strategy and the ability of expand and adaptive under this architecture.On a close study of scheduling algorithm in traditional grid job scheduling architecture based economy, this paper design an adaptive and dynamic grid job scheduling algorithm. This kind of algorithm dynamic monitor load balance degree of this system in the process of scheduling so that to get higher rate of success on job scheduling. Through our simulation trial prove that this kind of scheduling algorithm improve the scheduling success ratio of jobs compared with traditional DBC_COST algorithm in the same conditions.