I received my Ph.D. degree in Computer Science from the University of Southern California in May 2014. My doctoral research focused on benchmarking data stores. As a part of my dissertation, I developed the BG framework to evaluate the processing capability of data stores including their consistency, scalability, and performance characteristics. My motivation for developing BG was that the previous benchmarks focused either on simple OLTP or more complex OLAP operations without considering the requirements of social networking applications. These benchmarks exercised many aspects of data stores including throughput and response time but did not report on consistency issues and staleness measurements. As today’s social networking data stores focus on availability of data and relax the consistency requirements in favor of a higher availability, we developed BG to report on traditional performance metrics as well as the observed percentage of stale data. We also developed the BG framework to compute the Socialites and SoAR rating of data stores given a certain SLA requirements, allowing developers to evaluate the performance of their social networking applications using various data stores. BG was the basis of my Google 2013 PhD Fellowship award in Cloud Computing. My dissertation is the recipient of the 2014 best dissertation award from the USC Viterbi School of Engineering.
My research group has been investigating design and implementation of data intensive systems for more than two decades. The primary motivation for initiating BG is today's proliferation of many data stores and a scarcity of benchmarks to substantiate their claims. Sumita and I focused BG on interactive social networking actions because social networking companies continue to contribute data stores, e.g., TAO by Facebook, Voldemort by LinkedIn, and others. I am particularly interested to collaborate with social networking sites to certify an evaluation of their systems using BG (either in-house or a setup at the USC Database Laboratory).