Re: B-Heaps

Robert Haas wrote:
> This is drifting a bit off-topic for this thread, but it's not so easy
> to figure out from looking at the TODO which things are actually
> important. Performance-related improvements are mixed in with
> non-performance related improvements, which are mixed in with things
> that are probably not improvements at all. And even to the extent
> that you can identify the stuff that's performance-related, it's far
> from obvious which things are most important. Any thoughts on that

I don't think it's off topic at all actually, and as usually I'll be
happy to argue why. Reorganizing the TODO so that it's easier for
newcomers to consume is certainly a worthwhile but hard to "fund" (find
time to do relative to more important things) effort itself. My point
was more that statistically, *anything* on that list is likely a better
candidate for something to work on usefully than one of the random
theoretical performance improvements from research that pop on the lists
from time to time. People get excited about these papers and blog posts
sometimes, but the odds of those actually being in the critical path
where it represents a solution to a current PostgreSQL bottleneck is
dramatically lower than that you'll find one reading the list of *known*
issues. Want to improve PostgreSQL performance? Spend more time
reading the TODO, less looking around elsewhere for problems the
database may or may not have.

I have a major time sink I'm due to free myself from this week, and the
idea of providing some guidance for a "low hanging performance fruit"
section of the TODO is a good one I should take a look at. I have a
personal list of that sort already I should probably just make public,
since the ideas for improving things are not the valuable part I should
worry about keeping private anyway.

--
Greg Smith 2ndQuadrant US Baltimore, MD
PostgreSQL Training, Services and Support
greg(at)2ndQuadrant(dot)com www.2ndQuadrant.us