Enhancement in the neighbor-based diskless checkpointing approach

Abstract

In this paper we are providing a solution for the stable storage requirement in the disk-based system. Fault tolerant is very essential for distributed and parallel systems. To handle multiple processor failure here, we are using a diskless checkpointing approach. In that we are enhancing the neighbor-based approach. Instead of storing checkpoint in the stable storage we are storing it in peer processor’s main memory. To overcome the memory overhead we use a parity technique.