Analysis of data replication with two levels of consistency

A system employing data replication is structured as a two-level logical hierarchy. Replicas at level 0 maintain strong consistency, while those at level 1 may contain out-of-date information. Accesses which change the data have to be executed at level 0, and require an appropriate quorum of replicas. Read requests may be executed either at level 0 or at level 1, depending on whether the most recent information is required or not. A read quorum is necessary in the former case, but not in the latter. The performance and reliability characteristics of the system are evaluated by means of two models-one with reliable servers and one with random breakdowns and repairs. Much of the analysis is exact, but there are also some approximations.

[1]  Darrell D. E. Long.  Analysis of replication control protocols , 1990, [1990] Proceedings. Workshop on the Management of Replicated Data.

[2]  Peter J. B. King,et al.  Multiprocessor systems with preemptive priorities , 1981, Perform. Evaluation.

[3]  E. G. Coffman,et al.  A data base replication analysis using an M/M/m queue with service interruptions , 1982, SIGMETRICS '82.

[4]  Erol Gelenbe,et al.  Optimization of the Number of Copies in a Distributed Data Base , 1981, IEEE Trans. Software Eng..

[5]  P. Naor,et al.  Some Queuing Problems with the Service Station Subject to Breakdown , 1963 .

[6]  Darrell D. E. Long,et al.  On Improving the Availability of Replicated Files , 1987, SRDS.

[7]  Warren Smith,et al.  An Evaluation Method for Analysis of the Weighted Voting Algorithm for Maintaining Replicated Data , 1984, ICDCS.

[8]  Mostafa H. Ammar,et al.  Optimizing the performance of quorum consensus replica control protocols , 1990, [1990] Proceedings. Workshop on the Management of Replicated Data.

[9]  P. Triantafillou,et al.  High availability is not enough (distributed systems) , 1992, [1992 Proceedings] Second Workshop on the Management of Replicated Data.

[10]  Debasis Mitra,et al.  Probabilistic Models of Database Locking: Solutions, Computational Algorithms, and Asymptotics , 1984, JACM.

[11]  Vidyadhar G. Kulkarni,et al.  Stability and queueing time analysis of a reader-writer queue with alternating exhaustive priorities , 1995, Queueing Syst. Theory Appl..

[12]  Ram Chakka,et al.  Spectral Expansion Solution for a Class of Markov Models: Application and Comparison with the Matrix-Geometric Method , 1995, Perform. Evaluation.

[13]  Jean Bacon,et al.  A hierarchical asynchronous replication protocol for large scale systems , 1993, Proceedings 1993 IEEE Workshop on Advances in Parallel and Distributed Systems.

[14]  Vidyadhar G. Kulkarni,et al.  A reader-writer queue with reader preference , 1994, Queueing Syst. Theory Appl..

[15]  Mostafa H. Ammar,et al.  Performance Characterization of Quorum-Consensus Algorithms for Replicated Data , 1989, IEEE Trans. Software Eng..

[16]  David K. Gifford,et al.  Weighted voting for replicated data , 1979, SOSP '79.

[17]  E. Gelenbe,et al.  Optimization of the number of copies in a distribution data base , 1980, Performance.

[18]  Randolph D. Nelson,et al.  Analysis of a Replicated Data Base , 1985, Perform. Evaluation.

[19]  Jean Bacon,et al.  Performance evaluation of a hierarchical replication protocol: synchronous versus asynchronous , 1995, Second International Workshop on Services in Distributed and Networked Environments.

[20]  Costas Courcoubetis,et al.  Optimal control of a queueing system with simultaneous service requirements , 1987 .