On Mon, 2003-04-21 at 20:43, Tom Lane wrote:
> Jeremiah Jahn <jeremiah@cs.earlham.edu> writes:
> > The only issue with this is that it is difficult to recomend to our
> > clients who depend on bob and cuz'n joe to support their hardware.
>
> And you expect them to be successful running a database that acquires
> 2TB+ of data per year? I think you need to recalibrate your
> expectations. Consumer-grade junk PCs do not have the reliability
> to make such a project even worth starting. Run the database on decent
> made-to-be-a-server hardware, or you'll regret it.
That's the question...That 2 TB of data is nothing but documents and
images. I'm under the perception that if that gets parked on a fibre
channel disk array/ SAN the data will be pretty safe, and the server
mostly replaceable at that time. Storage is my worry more than
processing power. I don't think I'm on crack here...?
>
> I think I've spent more time chasing various people's hardware failures
> lately than I have in investigating real Postgres bugs. I keep
> volunteering to look at failures because I figure there are still some
> data-loss bugs to be found, but I am coming to have a *real* low opinion
> of off-the-shelf PC hardware.
>
> regards, tom lane
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 4: Don't 'kill -9' the postmaster
--
Jeremiah Jahn <jeremiah@cs.earlham.edu>