Skip site navigation (1) Skip section navigation (2)

Re: Reliability recommendations

From: Scott Marlowe <smarlowe(at)g2switchworks(dot)com>
To: "Craig A(dot) James" <cjames(at)modgraph-usa(dot)com>
Cc: "Joshua D(dot) Drake" <jd(at)commandprompt(dot)com>, pgsql-performance(at)postgresql(dot)org
Subject: Re: Reliability recommendations
Date: 2006-02-24 23:59:14
Message-ID: 1140825554.18756.26.camel@state.g2switchworks.com (view raw or flat)
Thread:
Lists: pgsql-performance
On Fri, 2006-02-24 at 17:12, Craig A. James wrote:
> Joshua D. Drake wrote:
> >> I find this strains credibility, that this major manufacturer of PC's 
> >> would do something deceptive that hurts performance, when it would be 
> >> easily detected and widely reported.  Can anyone cite a specific 
> >> instances where this has happened?  Such as, "I bought Dell model XYZ, 
> >> which was advertised to have these parts and these specs, but in fact 
> >> had these other parts and here are the actual specs."
> > 
> > I can :)
> > 
> > Feb 20 07:33:52 master kernel: [4294682.803000]   Vendor: MegaRAID  
> > Model: LD 0 RAID1   51G  Rev: 196T
> > --- snip ---
> > This machine... if you run it in raid 5 will only get 7-9 megabytes a 
> > second READ! performance. That is with 6 SCSI drives.
> > If you run it in RAID 10 you get a more reasonable 50-55 megabytes per 
> > second.
> 
> But you don't say how this machine was advertised.  Are there components in that list that were not as advertised?  Was the machine advertised as capable of RAID 5?  Were performance figures published for RAID 5?
> 
> If Dell advertised that the machine could do what you asked, then you're right -- they screwed you.  But if it was designed for and advertised to a different market, then I've made my point: People are blaming Dell for something that's not their fault.

IT was advertised as a rackmount server with dual processors and a RAID
controller with 6 drive bays.  I would expect such a machine to perform
well in both RAID 5 and RAID 1+0 configurations.

It certainly didn't do what we expected of a machine with the specs it
had.  For the same price, form factor and basic setup, i.e. dual P-IV 2
to 4 gig ram, 5 or 6 drive bays, I'd expect the same thing.  They were
crap.  Honestly.  Did you see the post where I mentioned that under
heavy I/O load they lock up about once every month or so.  They all do,
every one I've ever seen.  Some take more time than others, but they all
eventually lock up while running.

I was pretty much agnostic as to which servers we bought at my last job,
until someone started ordering from Dell and we got 2600 series
machines.  No one should have to live with these things.

In response to

pgsql-performance by date

Next:From: Luke LonerganDate: 2006-02-25 00:06:57
Subject: Re: Reliability recommendations
Previous:From: Craig A. JamesDate: 2006-02-24 23:12:43
Subject: Re: Reliability recommendations

Privacy Policy | About PostgreSQL
Copyright © 1996-2014 The PostgreSQL Global Development Group