> Especially as, in repeated tests, PostgreSQL with persistence turned off
> is just as fast as the fastest nondurable NoSQL database. And it has a
> LOT more features.
An option to completely disable WAL for such use cases would make it a lot
faster, especially in the case of heavy concurrent writes.
> Now, while fsync=off and tmpfs for WAL more-or-less eliminate the IO for
> durability, they don't eliminate the CPU time.
Actually the WAL overhead is some CPU and lots of locking.
> Which means that a caching version of PostgreSQL could be even faster.
> To do that, we'd need to:
>
> a) Eliminate WAL logging entirely
> b) Eliminate checkpointing
> c) Turn off the background writer
> d) Have PostgreSQL refuse to restart after a crash and instead call an
> exteral script (for reprovisioning)
>
> Of the three above, (a) is the most difficult codewise.
Actually, it's pretty easy, look in xlog.c
