Re: Freebsd vs linux and hardware question

Rich,

Thanks. I'm familiar with these. They're more along the line of once you
have your hardware here's how to optimize it.

I'm more interested in this points in things that talk about the merit
of different hardware setups. Obviously it's a moving target.

Regards,

Dror

On Fri, Sep 19, 2003 at 11:13:27PM +0000, Rich Seiersen wrote:
> There are two docs here on hardware optimization, one by Bruce Momjian:
> http://techdocs.postgresql.org/
>
>
>
> Richard Seiersen
> rich67dev(at)hotmail(dot)com
>
>
>
>
>
> >From: Dror Matalon <dror(at)zapatec(dot)com>
> >To: Josh Berkus <josh(at)agliodbs(dot)com>
> >CC: sfpug(at)postgresql(dot)org
> >Subject: Re: [sfpug] Freebsd vs linux and hardware question
> >Date: Fri, 19 Sep 2003 15:48:35 -0700
> >
> >On Fri, Sep 19, 2003 at 03:10:48PM -0700, Josh Berkus wrote:
> >> Dror,
> >>
> >> Howdy! Long time, no hear.
> >
> >Busy, busy as usal :-).
> >
> >>
> >> > Any opinions on the advantages/disadvantages of running postgres on
> >> > Linux vs running it on freebsd?
> >>
> >> Several prominent members of the PostgreSQL community claim better
> >performance
> >> on BSD due to BSD's better I/O stack, plus the history of Ingres
> >filesystem
> >> optimization on BSD. However, nobody has published figures.
> >
> >That's what I thought, but I wasn't sure if recent improvements in linux
> >IO hasn't caused it to catch up. I know that Linux networking has
> >improved a lot in the last couple of years, but I wasn't sure if it and
> >the IO code has caught up with BSD.
> >
> >>
> >> > Also, under either OS would a dual processor machine run faster than a
> >> > single processor one. For the most part, I know that databases are IO
> >> > bound rather than CPU, bound so more memory and more and faster disks
> >> > are a main issue. Still, it seems like being able to use multiple cpus
> >> > could improve performance.
> >>
> >> Not sure about BSD, but on Linux I know that Postgres will happily use
> >up to 4
> >> CPUs for simultaneos seperate requests. This doesn't help with modular
> >large
> >> queries/procedures, but does help a lot for concurrent users.
> >
> >OK.
> >
> >> >
> >> > Finally, is SCSI still such a big improvement over modern IDE drives?
> >>
> >> Yes. ;-)
> >>
> >> But hey, why are you asking this stuff? Don't you run Zapatec's stuff
> >off a
> >> huge SAN?
> >
> >It's not so Huge, out netapp is less than 200 Megs. But the real issue
> >is that it works really well for small data sets but not for big ones.
> >
> >We're working on a new application that will handle millions of rows.
> >Currently the table size is half a gig and growing.
> >
> >On this type of application the netapp is slower than local disks. So
> >we're thinking of setting up a new machine with plenty of power. I'm
> >trying to figure out what the right configuration would be both in terms
> >of the hardware and the OS.
> >
> >Any pointers to recommended hardware configurations?
> >
> >
> >Thanks,
> >
> >Dror
> >
> >
> >>
> >> --
> >> -Josh Berkus
> >> Aglio Database Solutions
> >> San Francisco
> >>
> >
> >--
> >Dror Matalon, President
> >Zapatec Inc
> >1700 MLK Way
> >Berkeley, CA 94709
> >http://www.zapatec.com
>
> _________________________________________________________________
> MSN 8 with e-mail virus protection service: 2 months FREE*
> http://join.msn.com/?page=features/virus
>

--
Dror Matalon, President
Zapatec Inc
1700 MLK Way
Berkeley, CA 94709
http://www.zapatec.com