Re: random_page_cost = 2.0 on Heroku Postgres

On Sat, Feb 11, 2012 at 8:26 AM, Jeff Janes <jeff(dot)janes(at)gmail(dot)com> wrote:
> By the time you realize the query is running long, it is too late to
> start analyzing it.  And without analyzing it, you probably can't get
> the information you need.
>
> Maybe with the timing = off feature,it would might make sense to just
> preemptively analyze everything.
>

As before, I'm reluctant to introduce structural performance costs
across the fleet, though I suppose giving users the option to opt out
might ameliorate that. I don't think I have time right now to
seriously explore these ideas, but I'll keep it in the back of my mind
for a while.

If anyone is interested in seriously exploring the idea of researching
query planner accuracy across an enormous fleet of production
databases with the goal of feeding that information back to the
project please feel free to contact me off-list.

--
Peter van Hardenberg
San Francisco, California
"Everything was beautiful, and nothing hurt." -- Kurt Vonnegut