> > In part 1. Differences in MVCC implementation - he's saying that "It’s > > not that the PostgreSQL implementation of MVCC is bad — it’s just > > fundamentally different" > It is written by someone @firebirdsql.org so one assumes a few grains of salt necessary. I know - but the guy does stress that he's not knocking PostgreSQL, just that there are differences. However, it *_was_* my understanding that MVCC was implemented similarly in PostgreSQL and Firebird - PG has VACUUM and FB has SWEEP. Why would FB need SWEEP if it didn't have to clear up after transactions - a problem that apparently doesn't affect Oracle/MySQL? Oracle and MySQL (InnoDB) implement a different model (as does Orioledb IIUC) where there's are UNDO/REDO logs. So, my question is: Is FB's MVCC implementation fundamentally different from that of PG or have I mixed things up? Thanks for your input. -- El!
