An experimental comparison of locking policies in a testbed database system

ABSTRACT

Multiuser transaction processing and database systems commonly use well-formed, two-phase locking to maintain data consistency. The preliminary results presented here represent the first step of an experimental investigation of the impact of different locking schemes on transaction performance. Tests were performed using a simplified but functionally complete "testbed" system. The transaction throughput rates for three different locking policies (file level locking, page level locking, and mixed level locking) are compared as a function of the lock mode (exclusive or share) and the file size for a small centralized database application. The experimental results for this environment show that the choice of locking policy and lock mode can have a significant impact on transaction throughput performance. The tests also demonstrate the sensitivity of the conclusions to the choice of workload and system characteristics.

REFERENCES

• 1: Philip A. Bernstein , Nathan Goodman, Concurrency Control in Distributed Database Systems, ACM Computing Surveys (CSUR), v.13 n.2, p.185-221, June 1981 [doi>10.1145/356842.356846]
• 2: {CHEN77} Chen, P.P. and Yao, S.B., "Design and Performance Tools for Data Base Systems," Proceedings, 3rd International Conference on Very Large Data Bases, 1977.
• 3: {CHUN81} Chung, M., "Design and Implementation of the Testbed Performance Monitor," Technical Report, Corporate Research Group, Digital Equipment Corporation, September 1981, 19 pages.
• 4: {GARC83} Garcia-Molina, H., F. Germano, and W. H. Kohler, "Architectural Overview of a Distributed Software Testbed," Proceedings Sixteenth Hawaii International Conference on System Sciences, January 1983.
• 5: {GERR76} Gerritsen, R., et al., "WAND User's Guide," Department of Decision Science, The Wharton School, University of Pennsyvania, April 15, 1976.
• 6: {GRAY76} Gray, J.N., R.A. Lorie, G.R. Putzolu, and I.L. Traiger, "Granularity of Locks and Degrees of Consistency in a Shared Data Base," in Modeling in Data Base Management Systems, Nijssen, editor, North Holland, 1976.
• 7: {GRAY79} Gray, J.N., "Notes on Data Base Operating Systems," in Operating Systems: An Advance Course, R. Bayer, R. H. Graham, and G. Seegmuller, Editors, Springer-Verlag, 1979, Sections 5.7.6 --- 5.7.7.3, pages 438 -- 450.
• 8: Keki B. Irani , Hing-Lung Lin, Queueing network models for concurrent transaction processing in a database system, Proceedings of the 1979 ACM SIGMOD international conference on Management of data, May 30-June 01, 1979, Boston, Massachusetts [doi>10.1145/582095.582116]
• 9: Walter H. Kohler, A Survey of Techniques for Synchronization and Recovery in Decentralized Computer Systems, ACM Computing Surveys (CSUR), v.13 n.2, p.149-183, June 1981 [doi>10.1145/356842.356845]
• 10: Daniel R. Ries , Michael Stonebraker, Effects of locking granularity in a database management system, ACM Transactions on Database Systems (TODS), v.2 n.3, p.233-246, Sept. 1977 [doi>10.1145/320557.320566]
• 11: Daniel R. Ries , Michael R. Stonebraker, Locking granularity revisited, ACM Transactions on Database Systems (TODS), v.4 n.2, p.210-227, June 1979 [doi>10.1145/320071.320078]
• 12: {RYPK79} Rypka, D.J., and A.P. Lucido, "Deadlock Detection and Avoidance for Shared Logical Resources," IEEE Transactions on Software Engineering, Vol. SE-5, No. 5, September 1979, pp. 465--471.