Social Web Database Benchmark

Social networking websites allow people to connect with others who share common interests, and provide a forum for interaction between them. They also allow people in these groups to keep track of what is happening in each others' lives.

The last few years have seen explosive growth not only in social network websites, but in applications which utilize data from these networks. Data stores and content on major social web applications are estimated to quadruple every 18 months, along with a doubling of queries per terabyte of data. This results in incredible pressure on web applications' database tiers, a pressure which will only increase in the years to come.

Database tiers at social networking sites see access patterns which are very different from typical OLTP and e-commerce applications. Whereas OLTP and e-commerce share a balanced read/write profile, social networking databases have very high read-to-write ratios. They also need to support hundreds of simultaneous user connections and majority of queries are point queries that return a few rows of a few hundred bytes, as compared to reporting queries which may scan entire databases and return hundreds of rows. Some database queries may also require joins or self joins for analysis of a graph, a particularly difficult to perform operation.