Re: PATCH: pgbench - merging transaction logs

> Guys, I don't see this theoretical discussion going anywhere. I think
> it's time to simply implement this and evaluate it on a bigger
> machine.

Sure. I was kind of hoping that someone else would implement it, because
I'm a reviewer on this one, and I do not have the bigger machine at hand
either:-) If nothing happens and when I have some time, I'll do it.

> It can't take very long to implement tosimply just write to one
> file instead of the multiple files as now.

Sure.

> The posix guaranteed fprintf locking should already take care of the
> rest.

I think I would try with snprintf to reduce locking.

Note that I'm first trying to evaluate logging overhead, but I'm quite far
from Tomas figures at the moment, looks more like 20% overhead (12 threads
read-only load, runs at about 130000 tps with full logging, 160000
without, repeated 5 times with the similar results).

In such a case, a mutex would be a bad idea, but I must add that detailed
logging is a plain bad idea as well!

--
Fabien.