Benchmark Cassandra and MongoDB with YCSB Using Different Amount of Shards, Replication Factors and Throughputs

• Main Authors: Chou, Li-Heng, 周立珩
• Other Authors: Wu, Tung-Kuang
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/ux6xy8

碩士 === 國立彰化師範大學 === 資訊管理學系所 === 105 === With the development of the information and communication technology, people can collect data easier than before. This evolution also speeds up the growth of the amount of data that can be used for analysis. This movement leads us into the big data era. However, due to the limitations of relational database, it can no longer meet the current big data requirements. Therefore, a new type of database model, NoSQL, has become the focus of various IT-related industries in this big data trend. When it comes to the choice of a specific NoSQL database, corporations must evaluate its characheristic with respect to their needs carefully. Accordingly, in this study we aim to conduct an all aspects evaluation of the top two NoSQL databases, MongoDB and Cassandra, so that related corporations may have better insight into these two systems. YCSB (Yahoo! Cloud Serving Benchmark) is adopted as the benchmarking program suit, to evaluate the two systems with different amont of shards, replication factors and throughputs. Our results show that Cassandra significantly outperformed MongoDB in most of our experiments.