Performance evaluation of locking protocol for distributed information processing

A distributed database is a decentralized system which distributes all its workloads to different working sites that are not tightly coupled. In a distributed database system, working data are grouped as records and exist redundantly in different locations. Shared data on processors can be remotely accessed through other connected nodes. As the data transmission technology rapidly develops, the transmission rate of communication networks has increased from a few Mb/s in coaxial cables to several Gb/s in fiber optical links. The performance of distributed database systems have been improved and extended. The major problem for replicated distributed database systems is complicated by the requirement to ensure mutual consistency for sharing resources. The two-phase locking algorithm is one common method to tackle this problem. This algorithm can ensure the system works properly without any deadlock problem. This paper simulates a two-phase locking algorithm and studies the effects for various workloads.