Abstract—In this paper some commonly used concurrency control protocols have been implemented through simulation. It is well known that the transactions have mainly four properties: atomicity, consistency, isolation and durability, which are known as the ACID properties. The objective of concurrency control is to ensure consistency when a shared database is updated by multiple concurrent transactions. It is also used to increase the database object utilization. Among all the existing concurrency control protocols, standards are locking, two phase locking, graph based protocols, time stamp-based protocol, wait-die scheme and wound-wait scheme. This paper analyze implementation of two phase locking, wait-die and wound-wait schemes. To implement these protocols, a PC based model has been developed first. The outcome of the model is quite satisfactory. The relative performance of those protocols has been compared in terms of utilization. For each protocol a concurrency control manager has been designated which maintains all lock buffers, detects deadlock, and takes necessary action for deadlock recovery. Analyzing the whole experiment it is clear that wait-die scheme and wound-wait scheme protocols show better performance than two phases locking.

Index Terms—Transaction, Serializability, Deadlock, Protocol

Samir Kumar Ghosh is with Dept. of Computer Engineering, National Institute of Technology Karnataka, Surathkal-575025, India.
Md. Shohidul Islam is with the Department of Computer Science & Engineering in Dhaka University of Engineering & Technology, Gazipur-1700, Bangladesh.
Md. Anisur Rahman completed graduation from Department of CSE in Rajshahi University of Engineering & Technology, Rajshahi-6204,Bangladesh.

Cite: Samir Kumar Ghosh, Md. Shohidul Islam, and Md. Anisur Rahman, "On Analytical Performance Measurement of Concurrency Control  Protocols  in  DBMS,"  International Journal  of  Computer  and  Electrical
Engineering vol. 1, no. 3, pp. 284-287, 2009.