Benchmarking cloud-based data management systems

Abstract

Cloud-based data management system is emerging as a scalable, fault tolerant and efficient solution to large scale data management. More and more companies are moving their data management applications from expensive, high-end ser-vers to the cloud which is composed of cheaper, commodity machines. The implementations of existing cloud-based data management systems represent a wide range of approaches, including storage architectures, data models, tradeoffs in consistency and availability, etc. Several benchmarks have been proposed to evaluate the performance. However, there were no reported studies about these benchmark results which provide users with insights on the impacts of different implementation approaches on the performance. We conducted comprehensive experiments of several representative cloud-based data management systems to explore relative performance of different implementation approaches the results are valuable for further research and development of cloud-based data management systems.