Study on Basics of Replica Control Management in Distributed Data

Abstract

A replicated database system is a distributed system in which each site stores a copy of the database (full replication) or parts of the database (partial replication). Data access is done via transactions. A transaction represents a logical unit of read and write operations. Two important components of a replicated database system are concurrency control and replica control. Concurrency control isolates concurrent transactions with conflicting operations, while replica control coordinates the access to the different copies. This Paper provides an informal introduction to database replication along with an overview of the advantages and disadvantages of traditional solutions. In particular, we elaborate on the main drawbacks of these solutions to be able to distinguish between inherent and avoidable limitations. This leads us to the key concepts behind the approach proposed in this dissertation which eliminate the avoidable and alleviate the inherent limitations of traditional solutions.