Application and middleware transparent checkpointing with TCKPT on clustergrid

Abstract

This paper introduces a way to transform the existing parallel checkpointing techniques to be applied for software-heterogeneous ClusterGrid infrastructures. While existing solutions are aiming at providing application transparency by building special middleware, this paper aims at targeting both application and middleware transparency at the same time by inserting checkpoint functionality into the application. The compatibility and integrity requirements are identified and corresponding conditions are established. Some of the available checkpointing systems are checked against the conditions in order to examine their conformity. Based on the conditions, a novel checkpointing method is defined and the TotalCheckpoint tool is adapted for ClusterGrid.