Adaptive persistence layer for synchronous replication (PLSR) in heterogeneous system / Abul Hashem Beg

Thesis (Master of Science (Computer)) -- Universiti Malaysia Pahang - 2011

Bibliography : p. 75-82

Nowadays, in the grid community, distributed and clustering system, a lot of work has been focused on providing efficient and safe replication management services through designing of algorithms and systems. For many reasons, businesses or specially enterprise business or industrial business use replication. Therefore, replication is a useful technique for distributed systems. It can improve the performance and the reliability of a database application. In addition, it can be considered as a data backup method in case of hardware failure, software corruption or even a natural disaster. A change of the main database is reflected, forwarded and applied at each of the replicated server which might be in a remote location. Replication in the heterogeneous system is a very promising and challenging platform which is a compound of multi environment. Proper mechanism is significantly required in order to manage the complex heterogeneous data replication. In this research, Persistence Layer for Synchronous Replication (PLSR) has been proposed to support heterogeneous systems. The main objective of this technique is to develop an adaptive persistence layer which consisted of reliable and smooth replication. This technique also introduces a multi thread based persistence layer, which supports early binding and parallel connection to the servers. All the replication servers established its connection through interfaces. Furthermore, similar with the Service Oriented Architecture (SOA) and the structure is flexible enough to modify i.e.; adding and removing replication server. The PLSR is proposed based on the multithreading technique in order to avoid the dependency of replicated server from the main server and to make the enterprise software more enhanced so that the system will never be unstable during system up-gradation or system crashes. Consequently,the implementation of this technique will be applicable to enterprise application such as bank, insurance, group of companies as well as a small and medium organization such as NGO. The new replication process will also be used in e-commerce application to secure user transaction information. The motivation of implementation is to make sure the data replication is easy to maintain and cost effective. The PLSR architecture model, workflow and algorithms are described. The PLSR has been developed using Java Programming language. The system requirements also have been elaborated. The experimental main server and replication servers were established in Windows and Linux platform using the local area network (LAN). Finally, series of experiments have been carried out by using different servers. The snapshot of implementation showed that the proposed framework works successfully with replicating data in different operating systems. The result shows that PLSR performs outstandingly and the value is 83.2 % and 2.49% than SQL server for transactional insert and synchronization in compare to time (seconds).