A Clustering Based Vertical Fragmentation and Allocation of a Distributed Database

Distributed Database Management System (DDBMS) is often surrounded with an issue of identifying the best design strategy. Efficiency of this design strategy to a great extent rely upon fragmentation of a relation. It additionally relies upon allocation of the fragment pieces to some sites within the network. Fragmentation and allocation are considered independent, even though they utilize similar information to accomplish a common target. Among the existing fragmentation strategies, vertical fragmentation is regularly viewed as an entangled fragmentation strategies than other on the grounds that the immense number of choices makes it about difficult to get an ideal answer for the vertical fragmentation issue. Allocation of a fragment is a NP-hard problem as it is complex to search an optimal site for holding the fragment. In this manner, we can just hope to discover a heuristic arrangement of solution for all these activities. The main purpose of this research is to present a clustering based fragmentation technique in which table attributes with similarity reside in same fragment cluster. As per the frequency of user query, affinity matrix is generated. This affinity matrix is further utilized to generate affinity cluster. Based on this affinity cluster, attributes having least Euclidian distance between them are considered as similar attributes and vertical fragments are obtained. After the fragments are generated those fragments are allocated to certain desired sites based on allocation table. This approach aims to solve the issue and complexity that occurs in previous vertical fragmentation approach.