A Novel Technique for Predicting the Optimum Time for Reallocating Fragments in MCRUD Based DDBS using Machine Learning

Abstract

Distributed processing is a way of improving the performance of database management systems. The total process of distribution of data involves fragmentation, replication, and allocation of the data in the system. MCRUD Matrix-based Fragmentation (MMF) introduces a way of partitioning the tables in the initial stage when there are no empirical data generated. This technique undoubtedly increased the performance of Distributed Database significantly, but it is not said when the optimum time for reallocating the data to increase the performance more based on empirical data is. In this paper, we have tried to find the optimum time for re-allocating the data based on empirical knowledge. We built a distributed database system using MCRUD Matrix-based Fragmentation (MMF) and later we generated some empirical data and reviewed the overall performance of the system. Then we made a data size and time required to re-allocate trade-off regarding when it will be the best time to re-allocate the data to gain better performance. The finding of our paper will be a breakthrough for whoever will use MCRUD Matrix-based Fragmentation to implement their distributed system. It will help them understand the optimum time for re-allocating their data to have better performance from the system.