FEW PERFORMANCE ISSUES IN DISTRIBUTED REAL TIME DATABASE SYSTEMS

Abstract

newline newlineDistributed Real Time Database System can be defined as database system that depends on real time data. Recent advances in computer technology, high speed network, high capacity storage and innovative sensors make us possible the alias image processing capabilities in distributed real time database system spread over multiple sites and connected together via communication channel. Most common method which is used to improve the performance of data access in distributed system is replication. A replicated database environment is composed of number of databases distributed at different sites for or by an individual site. To achieve the desired goal of data replication process, replication should be considered in accordance with certain strategies that organize and manage data. By keeping the problem discussed above in mind we have developed dynamic replication algorithm for replica selection, replica placement and replica updating. newline newlineThe performance of database depends not only on database architecture and algorithms but also on the requests served within the time limit. In this environment, new policies must be designed efficiently to handle the execution of transactions. Hence to develop efficient heuristic to minimize missed deadline transactions, we have proposed two heuristics: Number of Locks and Cohort Priority and Virtual Priority Assignment Policy to transactions, which allow fixing the priority of transactions and cohorts as well as deadline of cohorts. The main objective of buffer cache management is to reduce the number of disk accesses by keeping popular pages in buffer cache, thereby reducing the average time of disk pages. As we know that in distributed transaction processing hundreds of thousands of transaction instances every hour with some tens of millions of data accesses. So for efficient accessing of transactions pages we have developed new algorithm and implemented to observe the behaviour of our algorithm through simulation result. newline newline newline