The social network's latest reveal is about LinkBench, a new database benchmark for the Social Graph, which is being released this week on GitHub.
(For reference, according to Facebook engineers, the Social Graph is a graph data structure involving an "immense map" of people, places, and things and the connections between them.)
Touted as a tool for developers who need to benchmark and fine tune database systems, LinkBench was designed to replicate the data model, graph structure, and request mix of Facebook's MySQL social graph workload.
Tim Armstrong, a PhD candidate in computer science at the University of Chicago and previously an intern on Facebook's database engineering team, asserted in a blog post on Monday that LinkBench is a graph-serving benchmark -- not a graph-processing benchmark.
Armstrong explained that the difference is the former description "simulates the transactional workload from an interactive social network service while the latter simulates an analytics workload."
We also believe that the broader community working on databases and social applications can benefit from a realistic benchmark for storage and retrieval of social network and other graph-structured data. These applications place many unique demands on database infrastructure due to rapid growth, large volumes of data, and rich data models, yet there are few benchmarks that test performance for these workloads.