A Modified IDE Algorithm for Improving Transaction Scheduling Performance in Real-Time Databases

Abstract

Transaction scheduling for databases consists of a series of interconnected tasks, which aim at optimizing query performance. This performance is optimized by effective scheduling of queries, in such a way that query dependencies are quickly resolved, execution of queries must have better cache hits, optimum buffer utilization is done, and effective roll-back mechanisms are followed. A large number of systems are developed for achieving this task, and each of them has their individual nuances. These systems either have better cache hit ratios, or good buffer utilization, or have optimum query resolution capabilities. But integration of all these capabilities into a single scheduling solution is difficult to achieve, thereby various optimization algorithms are developed by researchers. In this work, a similar algorithm is developed which is based on a modified version of the standard improved differential evolution (IDE) algorithm. The algorithm aims at minimizing different redundancies in the transaction scheduling process, thereby improving overall query performance. Results of the proposed model are evaluated and compared with standard transaction processing models, and an efficiency improvement of 10% in terms of query execution delay & 15% in terms of average server response time is achieved. Further extensions to this work are also recommended in this text.