Reliability in three-tier systems without application server coordination and persistent message queues

ABSTRACT

When dealing with fault tolerance in three-tier systems, two major problems need to be addressed, that is how to prevent duplicate transaction executions when classical timeout based retransmission logics are employed, and how to ensure the agreement among the back-end databases despite failures (a transaction needs to be aborted or committed at all the involved databases independently of the failure scenario). In this paper we address these problems by proposing a fault tolerant protocol that, unlike previous solutions, (i) avoids the additional phase of storing the client request into a persistent message queue and (ii) avoids explicit coordination of middle tier application servers (during both normal behavior and fail-over). Our protocol reduces therefore the overhead imposed on the end-to-end interaction, thus improving user perceived responsiveness, and provides better scalability.

REFERENCES

• 1: Philip A. Bernstein , Meichun Hsu , Bruce Mann, Implementing recoverable requests using queues, Proceedings of the 1990 ACM SIGMOD international conference on Management of data, p.112-122, May 23-26, 1990, Atlantic City, New Jersey, United States
• 2: Philip Bernstein , Eric Newcomer, Principles of transaction processing: for the systems professional, Morgan Kaufmann Publishers Inc., San Francisco, CA, 1996
• 3: Svend Frlund , Rachid Guerraoui, A Pragmatic Implementation of e-Transactions, Proceedings of the 19th IEEE Symposium on Reliable Distributed Systems (SRDS'00), p.186, October 16-18, 2000
• 4: Svend Frølund , Rachid Guerraoui, Implementing E-Transactions with Asynchronous Replication, IEEE Transactions on Parallel and Distributed Systems, v.12 n.2, p.133-146, February 2001 [doi>10.1109/71.910869]
• 5: Svend Frølund , Rachid Guerraoui, e-Transactions: End-to-End Reliability for Three-Tier Architectures, IEEE Transactions on Software Engineering, v.28 n.4, p.378-395, April 2002 [doi>10.1109/TSE.2002.995430]
• 6: Idit Keidar , Sergio Rajsbaum, On the cost of fault-tolerant consensus when there are no faults: preliminary version, ACM SIGACT News, v.32 n.2, June 2001 [doi>10.1145/504192.504195]
• 7: M. Little and S. Shrivastava. Integrating the object transaction service with the Web. In Proc. of the 2nd Int. Workshop on Enterprise Distributed Object Computing (EDOC), pages 194--205. IEEE Computer Society Press, 1998.
• 8: P. Romano, F. Quaglia, and B. Ciciani. Ensuring e-Transaction through a lightweight protocol for centralized back-end database. In Proc. of the 2nd Int. Symposium on Parallel and Distributed Processing and Applications (ISPA), pages 903--913. LNCS, Springer-Verlang, 2004.
• 9: G. Shegalov, G. Weikum, R. Barga, and D. Lomet. EOS: Exactly-Once E-Service middleware. In Proc. of the 28th Conference on Very Large Databases (VLDB), pages 1043--1046. Morgan Kaufmann, 2002.
• 10: The Open Group. Distributed TP: The XA+Specification Version 2. 1994.
• 11: Transaction Processing Performance Council. TPC Benchmark C, Standard Specification, Revision 5.1. 2002.

INDEX TERMS

Primary Classification:
  C. Computer Systems Organization
  C.4 PERFORMANCE OF SYSTEMS
      Subjects: Fault tolerance

Additional Classification:
  H. Information Systems
  H.2 DATABASE MANAGEMENT
      H.2.4 Systems
          Subjects: Distributed databases

General Terms:
Performance, Reliability

Keywords:
distributed protocols, transaction processing