Abstract:
An efficient way of improving pelformance of a database management system is
distributed processing. Distribution of data involves fragmentation, replication and
allocation process. Previous research works provided fragmentation solution based on
empirical data about the type and frequency of the queries. These solutions are not
suitable at th~ .initial stage of a distributed database.
In this thesis we have presented a fragmentation technique namely MCRUD Matrix
based Fragmentation (MMF) that can be applied at the initial stage as well as in later
stages of a distributed database system for partitioning the relations. Instead of using
empirical data, we have developed the matrix namely Modified Create, Read, Update
and Delete to make fragmentation decisions properly. The main 'concept of MMF is
finding the precedence of attributes to increase data locality. We have named it
Attribute Locality Precedence (ALP). The r~lations have been fragmented considering
the highest ALP value an10ng the attributes
Allocation of fragments is done simultaneously in our technique. So using MMF, no
additional complexity is added for allocating the fragments to the sites of a distributed
database as fragmentation is synchronized with allocation. Performance of a DDBMS
can be improved significantly by avoiding frequent remote access and high data
transfer among the sites. Result shows that the proposed technique can solve initial
fragmentation problem of distributed system properly.
