Comparison of new generation cloud data storage systems withYCSB framework

NoSQL database systems have been becoming commonly used data platform for big data applications. It is necessary to analyze this system's weak and strong sides by using workload with low and high pay load in the light of specific parameters to make best selection among different NoSQL solutions. The purpose of this thesis is to analyze that weakness and strengths of each database system's under low work load by launching three different NoSQL database systems on Windows operating system and testing their abilities under low work load. In our study, six different work loads of Yahoo Cloud Service Benchmark (YCSB), a platform designed by Yahoo were utilized in order to test performances of NoSQL databases. Moreover, widely used in storage and management of big data, these two NoSQL systems 'performances' analyses of new abilities which they gained with the recent developments have been demonstrated by using the current versions of the MongoDB and Cassandra database systems. Therefore, we have provided the opportunity to compare the analysis and test results of old versions of systems such as MongoDB and Cassandra in the similar studies we have encountered in our literature researches with the test results in the new versions of our study. Furthermore, contributions of OrientDB becoming more popular in graph database world and going ahead to be alternative to Neo4J used and known most in this area were revealed by comparing MongoDB and Cassandra two systems which are flagship of NoSQL database systems. According to result obtained, the fact that MongoDB and OrientDB showed better performance in throughput executed under low workload in second than Cassandra and that OrientDB showed better performance in latency than MongoDB were observed. Cassandra was insufficient in terms of performance on Windows operating system with regard to these two database systems.