BG is a benchmark that rates a data store for processing interactive social networking actions using a pre-specified service level agreement, SLA. An example SLA may require 95% of issued requests to observe a response time faster than 100 milliseconds. BG computes two different ratings named SoAR and Socialites. In addition, it elevates the amount of unpredictable data produced by a data store to a first class metric, including it as a key component of the SLA and quantifying it as a part of the benchmarking process.

One may use BG for a variety of purposes ranging from comparing different data stores with one another, evaluating alternative physical data organization techniques given a data store, quantifying the performance characteristics of a data store in the presence of failures (either CP or AP in CAP theorem), among others. This study illustrates BG’s first use case, comparing a document store with an industrial strength relational database management system (RDBMS) deployed either in stand alone mode or augmented with memcached. No one system is superior for all BG actions. However, when considering a mix of actions, the memcached augmented RDBMS produces higher ratings.