Re: Postgresql on SAN - Mailing list pgsql-hackers

From Greg Stark
Subject Re: Postgresql on SAN
Date
Msg-id 87eknosjol.fsf@stark.xeocode.com
Whole thread Raw
In response to Re: Postgresql on SAN  (Simon Riggs <simon@2ndquadrant.com>)
Responses Re: Postgresql on SAN  (Yannick Lecaillez <yl@itioweb.com>)
List pgsql-hackers
Simon Riggs <simon@2ndquadrant.com> writes:

> On Tue, 2004-07-06 at 18:22, Yannick Lecaillez wrote:
>
> > I would have the pgsql-hackers genius for do that :) . I think its the
> > only feature which force company to buy 50000$ Oracle licence ...

Fwiw, I think you've underestimated the price on those Oracle licenses by an
order of magnitude at least.

If there are as many companies willing to pony up for some postgres developers
I'm sure there would be people interested, but it's not the kind of project
someone's going to be doing in their spare time. As Oracle found, it's *hard*.
And moreover, it results in a system that's hard to use. Those companies that
need are also ponying up much more than $50k/year just for the DBAs capable of
running such beasts.

Free Software runs on a very different operating model than commercial
software. Instead of a sharp division between paying clients and profiting
developers, most Free Software exists because the programmers themselves found
they had a need and solved it for themselves.

For that reason I would be skeptical about seeing huge clustered postgres
systems a la Oracle OPS, simply because it's a very specialized need, and not
one that any postgres developer is likely to run into on his own. They're more
likely to run screaming when asked to provide such a monster than sit down and
start coding...

What most people need is some way to promise rapid recovery from failures. In
my personal opinion the smoothest most reliable method of providing that is a
PITR-based warm standby machine. I'm overjoyed that someone else saw the same
need and has been working feverishly on that for 7.5. 

There does seem to be an awful lot of people on this list lobbying for some
feature or another. It always seems a bit weird, like a basic misunderstanding
is at play. The developers are working for their employers or for themselves.
It doesn't really matter how many new users the Windows port will bring on,
for example. This isn't some proselytising religion. It'll get done if a
developer needs it either for him- or herself or for a client, not because you
made some convincing argument about how there are lots of other people who
would benefit.

-- 
greg



pgsql-hackers by date:

Previous
From: Justin Clift
Date:
Subject: Loadable Oracle Personality: WAS "LinuxTag wrapup" thread
Next
From: "Greg Sabino Mullane"
Date:
Subject: Re: [Re] Re: PREPARE and transactions