Checkpoint and Rollback in Asynchronous Distributed Systems(<特集>Special Issue on Generation Database Technology for Internet, Multimedia and Mobile computing)

概要

論文の詳細を見る
This paper proposes novel protocols for taking checkpoints and rolling back the processes for recovery in asynchronous distributed systems. The checkpointing protocol is a kind of the communication-induced one. No additional message is transmitted for taking checkpoints. The rollback protocol allows processes to be restarted from checkpoints asynchronously. That is, each process can be restarted independently of the other processes. By using these protocols, livelocks which may be caused by the asynchronous restart can be avoided. The number of messages for a rollback is only O(l) where l is the number of channels. Hence, the system is highly available by the protocol presented in this paper.
一般社団法人情報処理学会の論文
1998-04-15