Proof and Evaluation of a 1CS Middleware Data Replication Protocol Based on O2PL
Middleware data replication techniques are a way to increase performance and fault tolerance without modifying the internals of a DBMS. However, they introduce overheads that may lead to poor response times. In this paper a modification of the O2PL protocol is introduced. It orders conflicting transactions by using their priority, instead of the total order obtained by an atomic multicast. Priorities are also used to avoid deadlocks. For improving its performance, it does not use the strict 2PC rule as O2PL does. We provide a formal correctness proof of its 1-Copy Serializability (1CS). This protocol has been implemented, and a comparison with other already implemented protocols is also given.
Unable to display preview. Download preview PDF.
- 1.Bernstein, P.A., Hadzilacos, V., Goodman, N.: Concurrency Control and Recovery in Database Systems. Addison Wesley, Reading (1987)Google Scholar
- 4.Jiménez-Peris, R., Patiño-Martínez, M., Kemme, B., Alonso, G.: Improving the scalability of fault-tolerant database clusters. In: ICDCS, pp. 477–484 (2002)Google Scholar
- 8.Armendáriz, J.E.: Design and Implementation of Database Replication Protocols in the MADIS Architecture. PhD thesis, Universidad Pública de Navarra, Pamplona, Spain (2006)Google Scholar