Heuristic Algorithm for Fragmentation and Allocation in Distributed Object Oriented Database

Abstract

Class fragmentation and allocation is an important technique to improve the performance of a distributed object oriented database system. The class fragmentation is to split a class into smaller pieces in distributed databases aims to reduce the access to unnecessary data, the allocation is to locate fragmented classes into the sites in the connected network properly to reduce the cost of data transmission. Classes in object databases include attributes describing the characteristics of the object, methods describing the behavior, and relationships with objects with other classes, including relations inheritance. With such characteristics, class fragmentation and allocation in the distributed object oriented database system is more complex than fragmentation technique and design of relational databases. Fragmentation techniques applied in the design of distributed object-oriented database today often do not use cost between the sites, fragments are allocated to the site after getting a fragmentation method of data objects. This paper proposes an algorithm of fragmentation and allocation simultaneously, including the cost of data communication between the sites used for fragmentation to reduce communication costs when processing and querying distributed data.