On Thu, 2006-05-18 at 17:10 -0500, Jim C. Nasby wrote:
> On Thu, May 18, 2006 at 10:31:03AM +0200, Martijn van Oosterhout wrote:
> > Besides, we still havn't got any reports yet that this actually
> > provides a benefit on any machine less than five years ago. Anyone out
> > there doing tests?
>
> Yes. I'm compiling the patched binaries right now, but the baseline
> testing I've got so far is at
> http://jim.nasby.net/misc/compress_sort.txt.
Looks a very good improvement. Well done Martijn/Jim.
The next question is: does it apply in all cases?
We need to test "SELECT aid from accounts" also, or some other scenarios
where the data is as uncompressible as possible. We should also try this
on a table where the rows have been inserted by different transactions,
so that the xmin value isn't the same for all tuples. We need to see if
there are cases where this causes a performance regression rather than
gain.
We still need to examine memory usage. Jim's testing so far is done on
already sorted data, so only uses 1 out of 715 tapes. If we did utilise
a much larger number of tapes, we could face difficulties with the
memory used during decompression.
--
Simon Riggs
EnterpriseDB http://www.enterprisedb.com