A Scalable Grid Scheduler for Real-Time Applications
Cong Liu, Sanjeev Baskiyar · Int. J. Comput. Their Appl. · 2009
In a grid computing environment, dynamicity and geographically distributed sites, make task scheduling problems challenging to solve. It is hard for a site to obtain precise real-time information about other sites whose load and computing resources may change dynamically. Moreover, the large number of individual resources of grid platforms raises scalability issues. Furthermore, many large scale data intensive applications make scheduling even more challenging since data transfer cost resources must be taken into consideration. In this paper we propose an innovative Peer-to-Peer Grid Scheduler (P2PGS) to solve such problems. P2PGS introduces two-phase scheduling framework that considers the large scale nature of the grid environment. P2PGS is composed of two phases: scheduling and dispatching. P2PGS is distributed and thus scalable since schedulers are distributed among resource sites. The scheduler at each site only needs the basic information of its neighbors to make dispatching decisions. Simulation results show that P2PGS can successfully schedule approximately 90 percent of computation-intensive tasks. For other tasks such as tasks with large data sets, it can successfully schedule more than 75 percent of them.