Transaction scheduling protocols for controlling priority inversion: A review

In advanced real-time distributed computing databases, the main performance criterion is to reduce the 'deadline miss' by the transactions; of course, consistency constraints also need to be satisfied. The goal of these applications is not to provide simply real-time transaction execution, but rather to provide a highly predictable, analysable, schedulable and reliable distributed computing platform to the users. The problem of resource conflicts amongst distributed real-time transactions and their handlings through various priority scheduling protocols highly affect the performance of the underlying applications. The past research works were mostly restricted to extend the traditional transaction processing techniques to resolve the issue of conflicts, and thus to improve the performance. The last review paper, largely on this issue, appeared in Shanker et al. (2008) [1]; since then many noteworthy algorithms have been described in the literature. Till date, no study was found discussing transaction processing techniques with data conflict issue in focus. Hence, our objective is to comprehensively discuss the state-of-the-art transaction scheduling protocols with an emphasis on the handling of execute-execute & execute-commit conflicts, and real-time optimistic concurrency control (OCC) protocols. The strengths and weaknesses of existing approaches are also discussed. (C) 2019 Elsevier Inc. All rights reserved.