Re: increasing collapse_limits?

-----BEGIN PGP SIGNED MESSAGE-----
Hash: RIPEMD160

Joshua Berkus wrote:

> I'm not comfortable with increasing the default, yet. While folks on
> dedicated good hardware can handle a collapse of 10-12 joins, a lot
> of people are running PostgreSQL on VMs these days whose real CPU
> power is no better than a Pentium IV.

Really? First, I don't think that's true, the average CPU power
is much higher than that. Second, this sounds like the 'ol
"tune it for a toaster" trap where we never make improvements
to the defaults because someone, somewhere, might *gasp* use
Postgres on an underpowered server.

> Also, if you're doing OLTP queries on small tables, spending 20ms
> planning a query is unreasonably slow in a way it is not for a
> DW query.

Again, seriously? Do you have numbers to back that up?

I could see not going to 16 right away, but who would honestly have a
problem with going to 10? I agree with Tom, let's bump this up a
little bit and see what happens. My guess is that we won't see a
single post in which we advise people to drop it down from 10 to 8.
Personally, I'd like to see them go to 12, as that's the best sweet
spot I've seen in the field, but I'll take 10 first. :)

Tom Lane asked re setting to 10:
> Don't know how much difference that would make in the real world though.

I've seen a handful of cases that have benefitted from 10, but many
more* that benefitted from 12 (*okay, a larger handful anyway, it's not
like I have to adjust it too often).

- --
Greg Sabino Mullane greg(at)turnstep(dot)com
End Point Corporation http://www.endpoint.com/
PGP Key: 0x14964AC8 201105012153
http://biglumber.com/x/web?pk=2529DF6AB8F79407E94445B4BC9B906714964AC8
-----BEGIN PGP SIGNATURE-----

iEYEAREDAAYFAk2+DqsACgkQvJuQZxSWSshRfQCgzX5JlnCmKTndA7WcF/mt0Kpk
b30AoLKrVKMm0rbZNNhgVjt/Xne4NDpj
=0deF
-----END PGP SIGNATURE-----