Abstract
Over recent years, the curve of the importance of data replication has risen steeply owing to the fact that databases are increasingly deployed over clusters of different workstations over time. A variety of replication techniques have been introduced to the distributed systems field which, in this paper, are classified based on whether they have an unbalanced load between servers or not (classic and modern). Replication techniques from both categories can be enhanced by avoiding some of the challenges that are illustrated in detail in this paper. Moreover, this paper analyses replication techniques in each category by exploring their strengths and weaknesses as well as providing possible novel solutions that can diminish or eliminate these challenges and introduces a brief description of the Dynamic Object Ownership Distribution Protocol that aims at increasing throughput by increasing the rate of performing transactions locally in addition to viewing a promising preliminary results of its performance.