Designe and implementation of fast retrieval distribution Database mangement system

number: 
754
إنجليزية
department: 
Degree: 
Imprint: 
Computer Science
Author: 
Ammar Hussain Abood Al-Ammeri
Supervisor: 
Dr.Lamya'a H. Khalid
Dr.Venus W. Samawi
year: 
2002
Abstract:

Distributed Databases have recently become an important area of S? information processing, at which it is easy to foresee that their importance will rapidly grow. Distributed Databases eliminate many of the shortcomings of centralized databases and fit more naturally in the decentralized structure of many organizations. This work concerns with design and implementation' of Fast Retrieving Distributed Relational Database Management. The proposed system is designed as a Three-Tier Multiple-Client/Multiple-Server model using JDBC-RMI components. The Three-Tier model has the advantages of separating the security requirement from the RDBMS (using Proxy server). The role of the proxy server is to check users authentication and provides concurrency controls. The use of JDBC-RMI components provides the proposed system with: the capabilities of working with other DBMSs, and the ability of remote execution (so that the only role of the system clients is to provide a pretty interface for users, while the whole execution is done in the RDBMS server sides). The proposed database server (RDBMS) provides fast retrieving for any database tuple (compared with the existing database systems) without taking in account the size of the database. This is done by suggesting and implementing an indexing technique which is a combination of Hashing and B+-Tree schemes. The most important property of Hashing scheme that has been used by the proposed system is the scattering of the data into non-equal subsets. The collision problem of Hashing is solved by indexing all the tuples that conflict with a key value by using B+-Tree indexing scheme to provide faster accessing to database tuples. The number of nodes in all B+-Trees is reduced by replacing the replicated keys with a single occurrence of the key and a list of tuple pointers. The proposed RDBMS provides six SQL queries that are divided into two types: DDL, which includes CREATE and DESTORYqueries, and DML, which include INSERT, DELETE, UPDATE, and SELECT. The system also provides five different domains: INTEGER, REAL, CHARACTER, BOOLEAN, and STRING.