2016 5th International Conference on Reliability, Infocom Technologies and Optimization (Trends and Future Directions)

11 / 2016-06-27 13:36:33

Docker for Optimization of Cassandra NoSQL Deployments on Node Limited Clusters

Docker,Cassandra,Resource Optimization,Distributed Computing,Performance Evaluation

World of computing is witnessing an immense growth in all aspects. Increased usage of internet for communication and the growth of Social Networking led to the invention of NoSQL database technologies capable of handling large amount of structured or unstructured data. Apache Cassandra is such an open source distributed data management system designed to handle large amount of data and is now being widely deployed in production environments. One of the important factors developers look for is cost effective deployment of Cassandra clusters by improving the resource utilization of existing clusters without increasing the number of nodes. In the production environment, only one instance of Cassandra can be deployed on one node of a cluster which can lead to under-utilized resources, if high specification servers are used. The cost involved increases as we increase the number of nodes in the cluster. In this paper, we propose a method to deploy an M+n node Cassandra cluster on M servers using Docker containers which allows deployment of multiple Cassandra instances on a single system and prove how effectively it can improve the performance of Cassandra NoSQL database with improved resource utilization without scaling the database into more number of nodes.