Efficient Commit/Abort Procedures in the RelaX Distributed Transaction Layer

  • R. Schumann
  • M. Mock
Conference paper
Part of the Informatik-Fachberichte book series (INFORMATIK, volume 214)

Abstract

Transactions are especially valuable in distributed systems, since they isolate the programmer from the effects of both concurrency and failures. In implementing transactions at the system level, flexibility has to be introduced into the transaction concept. Especially, the premature release of objects has to be dealt with. To assure recoverability nevertheless, resulting dependencies between transactions are stored by the system in a distributed data structure called recovery graph. By partially replicating the recovery graph at the different sites, the complexity of the commit procedure is reduced and a chase procedure used to abort transactions can be derived which avoids infinite chasing. The information transfer needed for replication is almost for free because it can be piggybacked on messages sent anyway. The new commit/abort procedures will be used in the RelaX project (Reliable Distributed Applications Support on Unix), which carries on work done at GMD in the PROFEMO project on distributed transaction mechanisms.

Keywords

Assure Weinstein Coord Kroger 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. [1]
    Chang, Maxemchuck: Reliable Broadcast Protocols, ACM Transactions, Vol. 2, August 1984.Google Scholar
  2. [2]
    Eswaran On the Notions of Consistency and Predicate Locks, CACM, Vol. 19, No. 11, 1976.Google Scholar
  3. [3]
    Gray: Notes on database operating systems, LNCS Vol. 60, 1978.Google Scholar
  4. [4]
    Merlin, Randell: State Restoration in Distributed Systems, Proc. 8th FTCS, Toulouse, 1978.Google Scholar
  5. [5]
    Müller: Implementation of Nested Transactions in a Distributed System, PhD Thesis, UCLA, 1983.Google Scholar
  6. [6]
    Nett, Grosspietsch, Jungblut, Kaiser, Kröger, Lux, Speicher, Winnebeck: PROFEMO - Design and Implementation of a Fault Tolerant Distributed System Architecture, GMD - Studien, Nr. 100, Juni 1985.Google Scholar
  7. [7]
    Nett, Kröger, Kaiser: Implementing a General Error Recovery Mechanism in a Distributed Operating System, FTCS 16, Vienna, Austria,1986.Google Scholar
  8. [8]
    Schumann: Transaktions-Verwaltung in einem verteilten objektorientierten System, GMD- Studie 134, Birlinghoven, 1988.Google Scholar
  9. [9]
    Skeen: Nonblocking Commit Protocols, Proc. ACM SIGMOD Int. Conf. on Management of Data, 1981.Google Scholar
  10. [10]
    Spector: Camelot: A Distributed Transaction Facility for Mach and the Internet - An Interim Report, TR CMU-CS-87-129, Computer Science Department, Carnegie-Mellon-University, 1987.Google Scholar
  11. [11]
    Weinstein Transactions and Synchronisation in a Distributed Operating System, 10th ACM Symp. on Operating System Principles, 1985.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1989

Authors and Affiliations

  • R. Schumann
    • 1
  • M. Mock
    • 1
  1. 1.Gesellschaft für Mathematik und Datenverarbeitung (GMD) Schloß BirlinghovenGermany

Personalised recommendations