A Low Overhead Minimum Process Global Snapshop Collection Algorithm for Mobile Distributed System

摘要

Coordinated checkpointing is an effective fault tolerant technique indistributed system as it avoids the domino effect and require minimum storagerequirement. Most of the earlier coordinated checkpoint algorithms block theircomputation during checkpointing and forces minimum-process or non-blocking butforces all nodes to takes checkpoint even though many of them may not benecessary or non-blocking minimum-process but takes useless checkpoints orreduced useless checkpoint but has higher synchronization message overhead orhas high checkpoint request propagation time. Hence in mobile distributedsystems there is a great need of minimizing the number of communication messageand checkpointing overhead as it raise new issues such as mobility, lowbandwidth of wireless channels, frequently disconnections, limited batterypower and lack of reliable stable storage on mobile nodes. In this paper, wepropose a minimum-process coordinated checkpointing algorithm for mobiledistributed system where no useless checkpoints are taken, no blocking ofprocesses takes place and enforces a minimum-number of processes to takecheckpoints. Our algorithm imposes low memory and computation overheads on MH'sand low communication overheads on wireless channels. It avoids awakening of anMH if it is not required to take its checkpoint and has reduced latency time aseach process involved in a global checkpoint can forward its own decisiondirectly to the checkpoint initiator.