SCHEDULING IN VIRTUALIZED GRID ENVIRONMENT USING HYBRID APPROACH

ABSTRACT A typical grid application needs requirements in terms of memory, software and so on. Scheduling those applications and giving the results back to the client is the prime goal. In this paper, the scheduling algorithm is proposed for assigning jobs to the physical resources. When the software requirements of the application are not satisfied by the physical resource, the jobs are scheduled to a virtual machine. This virtual machine is created such that it satisfies the requirements of the job. This is the hybrid approach of scheduling proposed in this paper. The algorithm allocates dynamic priority to the jobs based on execution time and deadline and classifies them into multi-queues based on priority allocated. The bipartite graph is used for resource matching which gives the best resource match and reduces the conflicts between jobs for a particular resource. The approach in this paper is different as the dynamic user defined virtual environment is created only when there is no such physical environment already available. The reliability of the system is improved by proactive failure detection and RPC fault tolerance mechanism. The proposed system handles four categories of faults. Load balancing is also achieved through receiver initiated strategy