Quasi-dynamic two-phase locking

  • Authors:

  • Abdelsalam Helal;Tung-Hui Ku;Jud Fortner

  • Affiliations:

  • Computer Science and Engineering Department, The University of Texas at Arlington, Box 19015, Arlington, TX;Computer Science and Engineering Department, The University of Texas at Arlington, Box 19015, Arlington, TX;Computer Science and Engineering Department, The University of Texas at Arlington, Box 19015, Arlington, TX

  • Venue:
  • CIKM '94 Proceedings of the third international conference on Information and knowledge management
  • Year:
  • 1994

Quantified Score

Hi-index 0.00

Visualization

Abstract

Among the plethora of concurrency control algorithms that have been proposed and analyzed, two-phase locking (2PL) has been adapted as the industry de facto standard concurrency control. In accord, current research in concurrency control is focusing on enhancing the scalability of 2PL performance in highly concurrent and contentious environments. This is especially needed in future on-line transaction processing systems, where thousand Transaction Per Second performance will be required.Static locking (SL) and dynamic locking (DL) are two famous adaptations of 2PL that are used under different degrees of data contention. In this paper, we offer our observation that 2PL is indeed a family of methods, of which SL and DL are extreme case members. Further, we argue for and verify the existence of other 2PL member methods that, under variable conditions, outperform SL and DL. We propose two novel schemes which we categorize as quasi-dynamic two-phase locking on account of their behavior in comparison with dynamic/static two-phase locking. We present a simulation study of the performance of the proposed schemes and their comparison to dynamic and static locking methods.