ANALYSIS OF DEADLOCK DETECTION AND RESOLUTION TECHNIQUES IN DISTRIBUTED DATABASE ENVIRONMENT
Himanshi Grover, Suresh Kumar V · 2012
Deadlock is one of the most serious bottlenecks in multitasking concurrently running systems and it has become further complicated in distributed environment. Deadlock detection and their resolution is quite difficult in distributed systems as it involve data located at different sites. The deadlock problem is intrinsic to a distributed database systems which employs locking process in its concurrency control system. In literature various techniques have been discussed and used to prevent, detect and resolve the deadlocks. In this paper we have analyzed some deadlock detection and resolution techniques used. We have reviewed in detail the algorithms presented by B. M. Alom and Brian M. Jhonston for deadlock detection and resolution in distributed environment and found that when the order of the transactions is changed in the algorithm by Aloms then it fails to detect deadlocks whereas algorithm by B. M. Jhonston aborts a transaction for which the intersection values of Wait-for and Request-Q values is not nil. We have introduced the concept of time stamping to avoid these deficiencies.