Vertical fragmentation and allocation in distributed databases using k-mean algorithm

Distributed database systems are increasingly becoming the dominant tools for data management. However, in these systems, the sites are remote and exchange a huge amount of data, which leads to bottlenecks as well as large disk accesses in data transfers that may be irrelevant. Query processing optimization techniques are an important concern for system administrators to improve the performance of distributed database systems (DDBS). Vertical fragmentation is a suitable solution but its complication lies in the large number of alternatives to obtain an optimal solution. This paper provides a new approach better suited to the problem of vertical fragmentation by the Kmeans classification algorithm but with our new adequate distance. To validate our approach, we compared our solution first with a vertical fragmentation algorithm called VFAR and second with the same k-means algorithm with the hamming distance.