Re: Set visibility map bit after HOT prune

On Thu, Dec 20, 2012 at 5:05 AM, Simon Riggs <simon(at)2ndquadrant(dot)com> wrote:
>
>
> And those last things are being debated hotly. We definitely need to
> ask whether the way things are now can be tweaked to be better. The
> major mechanics need not be reviewed, but the tradeoffs and balances?
> Definitely.
>

I have zero objection to do that, just that I don't have solid ideas
right now. And its not because I haven't thought hard enough.

> Anything we do in foreground needs evaluation. Assuming eager actions
> give a good payoff is not always a useful thought.
>

I don't disagree. Your field experience is much larger than mine, but
I have spent hours testing PostgreSQL's performance, so can talk with
some degree of conviction. I think when we do things that can reduce
read/write IO or bloat of a large table in general, the system as a
whole benefits, may be at a cost of some genuinely good guy doing a
simple SELECT in this case. Often the SELECTs are also benefited
because one of their good siblings helped us reduce bloat of the table
and hence seq scans had to scan order of magnitude less blocks.

I just thought that we can fairly easily limit the damage if we are
really worried about SELECTs being penalised. What if we set a
configurable limit on *extra* things that a query may do which is
otherwise not very useful for the query itself, but is useful to keep
the system healthy and steady. HOT prune definitely counts as one of
them and may be even setting of hint bits. (This is a topic for a
separate thread though)

Thanks,
Pavan

--
Pavan Deolasee
http://www.linkedin.com/in/pavandeolasee