So they added some bureaucracy and wrote an incremental diff system instead of using Oracle which allowed them to increase the number of machines handling traffic...
A long winded recruitment post with no technical details.
This is probably the first time I have heard of OrientDB being used in production. It seemed like a unique database but it wasn't backed by a big commercial entity. It is used as the main datastore and the edges cache this data?
Basically the data is moving from Oracle to OrientDB then the edges cache the thrift data on the filesystem and in-memory.