Re: Postgres as In-Memory Database?

[Stefan Keller <sfkeller@xxxxxxxxx>]

Hi Andrew

You wrote: 

> And indeed, given the specifics of the use
> case you're outlining, it's as much a demonstration of that evaluation
> as a repudiation of it.

Maybe my use cases seem to be a special case (to me and over a million users of OpenStreetMap it's not).

Anyhow: That's why I'm investigating an FDW extension.

> I don't think there's any evidence that the Postgres developers ignore
> useful optimisations.  What you're arguing is that the optimisation
> you have in mind isn't covered. 

No; my point is that I - and others like Stonebraker, Oracle and SAP etc. - see room for optimization because assumptions about HW changed. To me, that should be enough evidence to start thinking about enhancements.

Yours, S.

 

2013/11/19 Andrew Sullivan <ajs@xxxxxxxxxxxxxxx>

On Tue, Nov 19, 2013 at 02:39:17AM +0100, Stefan Keller wrote:
> Referring to the application is something you can always say - but
> shouldn't prevent on enhancing Postgres.

With respect, that sounds like a sideways version of, "You should
optimise for $usecase".  You could be right, but I think the judgement
of the Postgres developers has generally been that special cases are
not the mainline case.  And indeed, given the specifics of the use
case you're outlining, it's as much a demonstration of that evaluation
as a repudiation of it.

I don't think there's any evidence that the Postgres developers ignore
useful optimisations.  What you're arguing is that the optimisation
you have in mind isn't covered.  What you need is an argument that it
is generally useful.  Otherwise, the right thing to do is get a
specialised tool (which might be a special optimisation of the
Postgres code).

Best,

A

--
Andrew Sullivan
ajs@xxxxxxxxxxxxxxx

--
Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general