Open Access Thesis
Date of Award
Benchmarking is a common method in evaluating and choosing a NoSQL database. There are already lots of benchmarking reports available in internet and research papers. Most of the benchmark reports measure the database performance only by overall throughput and latency. This is an adequate performance analysis but need not to be the end. We define some new perspectives which also need to be considered during NoSQL performance analysis. We have demonstrated this approach by benchmarking HBase, MongoDB and sharded MySQL using YCSB. Based on the results we observe that NoSQL databases do not consider the capability of the data nodes while assigning data to it. And these databases' performance is seriously affected by the bottleneck nodes and the databases are not attempting to resolve this bottleneck situation automatically.
Balasubramanian, Harish, "Performance Analysis Of Scalable Sql And Nosql Databases : A Quantitative Approach" (2014). Wayne State University Theses. 327.