TAO: The power of the graph

After a history of performance improvements starting with Mark Zuckerberg putting memcache on Facebook's Apache servers in 2005:

"In early 2009, a team of Facebook infrastructure engineers started to work on TAO (“The Associations and Objects”). TAO has now been in production for several years. It runs on a large collection of geographically distributed server clusters. TAO serves thousands of data types and handles over a billion read requests and millions of write requests every second."

It is solving some impressive problems, and kudos to Facebook for continuing to share neat bits of engineering with us.

The interesting part is that they have not gone for a custom graph representation on disk like neo4j et al, but are persisting the graph data as MySQL records itself.

Some interesting files that go along with the talk/presentation:

https://www.usenix.org/sites/default/files/conference/protec...

https://www.usenix.org/system/files/conference/nsdi13/nsdi13...