Vertical Fragmentation using K-Medoid in Distributed Database

Fragmentation and allocation are major challenges that improve the performance of distributed databases meticulously. Many algorithms are presented in the literature for horizontal and vertical fragmentation. By considering the advantages of vertical fragmentation, this paper presents the k-medoid hard clustering technique which is used for non-overlapping vertical fragmentation. Dissimilarity of each attribute with respect to the medoid is calculated and least dissimilar attributes are placed together to form fragments. Dissimilarity Count is used as a measure to check the appropriateness of the fragments formed. The proposed algorithm provides optimal results for fragmentation as compared to Enhanced Minimum Spanning Tree (EMST).