Clustering with Terracotta

Abstract

In today’s java community, modern enterprise application products have more constraints and requirements then ever. High availability, application scalability and also good performance are required, which means an application is needed to be deployed on multiple JVMs, in other words, it has to be clustered or distributed. It is essential for the application to scale out well, has better performance and less complexity during development of clustering. This master thesis focuses on clustering with Terracotta which is a JVM level clustering technique. First I start analyzing the complexity when an application comes into scale-out, and also analyzing the shortcomings of the common approaches to clustering an application. Then I do a deep dive to the Terracotta and demonstrate how to utilize Terracotta to conquer these problems with emphasis on providing scalability and high performance using natural java programming. Finally, various scenarios are made as benchmark tests. The final result has shown that Terracotta as redundancy solution is strong recommended to be implemented for high availability and high scalability