On Sun, 5 Oct 2003, Neil Conway wrote:

> > I don't know any portable way to do that :-(
> For the non-portable way of doing this, are you referring to O_DIRECT?
> Even if it isn't available everywhere, it might be worth considering
> this at least for the platforms on which it is supported.

I strongly agree here only if we can prove there is a benefit.
I think it would be silly of us if some OS supported SnazzyFeatureC that
was able to speed up PG by a large percentage (hopefully, in a rather
non-invasive way in the code).  But, I do see the problem here with bloat
and PG being radically different platform to platform.  I suppose we could
dictate that at least N os's had to have it.. or perhaps supply it has
contrib/ patches.... Something to think about.

I'd be interested in tinkering with this, but I'm more interested at the
moment of why (with proof, not antecdotal) Solaris is so much slower than
Linux and what we cna do about this.  We're looking to move a rather large
Informix db to PG and ops has reservations about ditching Sun hardware.


---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
      joining column's datatypes do not match

Reply via email to