Twitter publishes its FlockDB

It's nice that many large companies continue the tradition of putting serious things from internal development into Open Source.
Recently Twitter posted FlockDB

"FlockDB is a database that stores data about graphs. At the same time, it is not a database optimized for graph traversal operations. FlockDB is optimized for working with very large lists of adjacent graph vertices , fast reading and writing, and page-wise arithmetic operations on queries ( page-able set arithmetic queries). "

The main objective of the project was to solve Twitter problems with working with very large graphs of user social data (followers, operations, mention , etc.). The migration ended 9 months ago and it seems they are happy with everything now.
')


Currently, the system stores 13 billion graph edges and supports 20 thousand write operations and 100 thousand read operations per second.
More details on the product can be found at the link .
Sources are available on github .