An adaptive replication model for heterogeneous systems

Data replication is an increasingly important topic as databases are more and more deployed over distributed systems, grid community and clustering systems. The performance, reliability and portability of entire database may possible by using replication technique. Replication may be considered as a data backup policy. Replication in homogeneous system is common practice in real life, but replication in heterogeneous system is quite challenging, because of the dissimilar computing environment. Since the computer environment porn to be heterogeneous, hence it's a promising field for researchers to consider replication in heterogeneous environment. In our research a persistence layer has been proposed for replication in heterogeneous systems. This persistence layer work on asynchronous model, hence it may call as asynchronous replication model. The model works implements multi threading technique for creating parallel connection with peer servers. The main server and replicated server are connected with a common interface. The interface is a replication engine, which intelligently holds data and makes decision depending on different factors for sending data to smoothen the replication process. The whole structure follows the rules of SOA (Service oriented architecture) thus, modification of replication servers do not affect the main server. Finally the architecture of this concept builds on different configurable files. These files help us in system up-gradation without shutting down the system. At the end, some experiments have been carried out and the results have been analyzed.

I. Introduction

Data grid mainly deals with large computational problems in large distributed database environment. It can support to manage geographically distributed resources for large-scale data-intensive applications that may generate huge scientific data sets. For such large-scale systems scientists need to managing the massive amounts of largely geographically distributed data [2]. The effective management of these large amount of information shared and distributed is becoming a topic of scientific research for commercial applications. In a nutshell, data replication is very useful for distributed database systems. Over the decade, many researchers have extensively worked to propose different replication models. Prior research shows that in heterogeneous system, data replication is challenging and an issue of research and the study was initiated.