FlockDB Description
A distributed, fault-tolerant graph database. FlockDB is a distributed graph data base for storing adjancency tables. It has the following goals: support for high rates of add/update/remove operations, potientially complicated set arithmetic query, paging through query results sets containing millions, ability to "archive and later restore archived edges", horizontal scaling, including replication, and online migration. Multi-hop queries (or graph walking queries) and automatic shardmigrations are non-goals. FlockDB is simpler than other graph databases like neo4j, as it attempts to solve fewer problems. It can scale horizontally and is optimized for low latency, high throughput environments like web-sites. FlockDB is used by Twitter to store social graphs (who follows who, who blocks whom) as well as secondary indices. The Twitter FlockDB cluster has 13+ billion edges as of April 2010 and can sustain peak traffic of 20k write/second and 100k readers/second.