Reducing Transaction Abort Rates with Prioritized Atomic Multicast Protocols

Priority atomic multicast is a total-order multicast message delivery service that enables applications to prioritize the sequence by which messages are delivered, while regular total order properties remain invariant. Priority-based message delivery can serve to reduce the abortion rate of transactions. In this study, we compare three classical total order protocols against their corresponding prioritized versions, in the framework of a replication middleware. To this end, we use a test application that broadcasts prioritized messages by these protocols, and measure the effect of the priorization. We show that, under certain conditions, the use of prioritized protocols yields lower abort rates than the corresponding non-prioritized protocols.

[1]  Paulo Veríssimo,et al.  Priority-based totally ordered multicast. , 1995 .

[2]  Andrew S. Tanenbaum,et al.  Computer Networks , 1981 .

[3]  Akihito Nakamura,et al.  Starvation-prevented priority-based total ordering broadcast protocol on high-speed single channel network , 1993, [1993] Proceedings The 2nd International Symposium on High Performance Distributed Computing.

[4]  Idit Keidar,et al.  Group communication specifications: a comprehensive study , 2001, CSUR.

[5]  Theodore P. Baker,et al.  Stack-based scheduling of realtime processes , 1991, Real-Time Systems.

[6]  Emmanuelle Anceaume,et al.  Avoiding priority inversion on the processing of requests by active replicated servers , 2001, 2001 International Conference on Dependable Systems and Networks.

[7]  Emili Miedes,et al.  Adding Priorities to Total Order Broadcast Protocols , .

[8]  Francesc D. Muñoz-Escoí,et al.  Managing Priorities in Atomic Multicast Protocols , 2008, 2008 Third International Conference on Availability, Reliability and Security.

[9]  Louise E. Moser,et al.  Totem: a fault-tolerant multicast group communication system , 1996, CACM.

[10]  Francisco Castro-Company,et al.  MADIS: A Slim Middleware for Database Replication , 2005, Euro-Par.

[11]  Gustavo Alonso,et al.  Database replication techniques: a three parameter classification , 2000, Proceedings 19th IEEE Symposium on Reliable Distributed Systems SRDS-2000.

[12]  Kenneth P. Birman,et al.  Reliable communication in the presence of failures , 1987, TOCS.

[13]  Yair Amir,et al.  A low latency, loss tolerant architecture and protocol for wide area group communication , 2000, Proceeding International Conference on Dependable Systems and Networks. DSN 2000.

[14]  Andrew S. Tanenbaum,et al.  An evaluation of the Amoeba group communication system , 1996, Proceedings of 16th International Conference on Distributed Computing Systems.

[15]  Danny Dolev,et al.  The Transis approach to high availability cluster communication , 1996, CACM.

[16]  Leslie Lamport,et al.  Time, clocks, and the ordering of events in a distributed system , 1978, CACM.

[17]  Akihito Nakamura,et al.  Priority-based total and semi-total ordering broadcast protocols , 1992, [1992] Proceedings of the 12th International Conference on Distributed Computing Systems.

[18]  Santosh K. Shrivastava,et al.  Preventing state divergence in replicated distributed programs , 1990, Proceedings Ninth Symposium on Reliable Distributed Systems.