Multi-cycle Deadlock Detection Algorithm for Distributed Systems

Abstract

Deadlock handling is an important component of transaction management in a database system. Though a lot of works have already done for deadlock detection on distributed system. This paper proposes a multi-cycle deadlock detection and recovery mechanism. Our proposed algorithm has modified the probe based distributed algorithm for deadlock detection such as CMH algorithm. But CMH algorithm had some limitation. It can only detect single cycle. But some situation a node is responsible for multi-cycle deadlock detection. In this situation, it can’t detect Multi-cycle deadlock. Besides it can only detect deadlock when the initiator node involved in the cycle. Another algorithm named MC2DR has worked on CMH algorithm. But their probe message contains four fields for deadlock detection which takes more space, but our algorithm has reduced the MC2DR algorithm probe message field and increased the efficiency of this algorithm. It can detect multi-cycle and also detect which node is responsible for multi-cycle deadlock and kills it.