Fault Tolerant Dynamic Scheduling in Grid Environment

A Fault-tolerant dynamic scheduler meant for grid applications that uses job replication along with combines the compensation of static scheduling, specifically no overhead for the mistake free case, and low overhead in case of a fault. The dependency arrangement of a parallel program can be described by means of a directed acyclic graph (DAG). Scheduling graph consists of ordering the jobs and mapping them to processing units. Faults can happen at any point during time. A fault can be detected either because a processing unit and it does not send data at a prearranged time. Provide that fault-tolerance can be achieved while detecting that a processing unit has stopped through completing of a job, the status of this job is recovered from some checkpoint, and this job is re-executed on another processing unit, and all jobs (potentially) to be executed on this processing unit are reorganize to one more processor. Co-allocation involves transmission jobs to favorable resources. This is achieved by obtaining information from a variety of nodes of the grid and updating them as and when a modification occurs and assigning resources to the jobs based on their specifications and performances.