Fault tolerance using "Parallel shadow image servers (PSIS)" in grid based computing environment

This paper will present a critical review, of the existing fault tolerance mechanism in grid computing and the overhead involved in terms of reprocessing or rescheduling of jobs, if in case a fault arisen For this purpose we suggested the Parallel Shadow Image Server (PSIS) copying techniques in parallel to the Resource Manager for having the check points for rescheduling of jobs from the nearest flag, if in case the fault is detected. The job process is to be scheduled from the resource manager node to the worker nodes and then its' submitted back by the worker nodes in serialized form to the Parallel Shadow Image Servers from the worker nodes after the pre-specified amount of time, which we call the recent spawn or the flag check point for rescheduling or reprocessing of job. If the fault is arisen then the rescheduling will be done from the recent check point and will be submitted to the worker rode from where the job was terminated. This will not only save time but will improve the performance up to major extent.