Divergence Control Algorithms for Epsilon Serializability

  • Authors:

  • Kun-Lung Wu;Philip S. Yu;Calton Pu

  • Affiliations:

  • -;-;-

  • Venue:
  • IEEE Transactions on Knowledge and Data Engineering
  • Year:
  • 1997

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents divergence control methods for epsilon serializability (ESR) in centralized databases. ESR alleviates the strictness of serializability (SR) in transaction processing by allowing for limited inconsistency. The bounded inconsistency is automatically maintained by divergence control (DC) methods in a way similar to SR is maintained by concurrency control (CC) mechanisms. However, DC for ESR allows more concurrency than CC for SR. We first demonstrate the feasibility of ESR by showing the design of three representative DC methods: two-phase locking, timestamp ordering and optimistic approaches. DC methods are designed by systematically enhancing CC algorithms in two stages: extension and relaxation. In the extension stage, a CC algorithm is analyzed to locate the places where it identifies non-SR conflicts of database operations. In the relaxation stage, the non-SR conflicts are relaxed to allow for controlled inconsistency. We then demonstrate the applicability of ESR by presenting the design of DC methods using other most known inconsistency specifications, such as absolute value, age and total number of nonserializably read data items. In addition, we present a performance study using an optimistic divergence control algorithm as an example to show that a substantial improvement in concurrency can be achieved in ESR by allowing for a small amount of inconsistency.