Performance 4.x vs. 6.x

[Mike Horwath]

On Sat, Oct 14, 2006 at 01:13:27PM -0300, NOC Prowip wrote:
> On Saturday 14 October 2006 12:38, Mike Horwath wrote:
> > On Sat, Oct 14, 2006 at 11:13:24AM -0300, NOC Prowip wrote:
> > > Hi, I am hooking in here without any intention to fire things up but
> > > isn 't this discussion certainly useless? Not only 4.11 is gone but
> > > also i386 is practically marked to die out as well as UP systems
> > > are.
> >
> > Wow, I hope not.
> 
> only a matter of time I guess, next year we will have 64bit
> quad-cores and I am really not sure if anybody will build 32bit
> versions ever again

Again, I hope not.

> > Unless you are separating out i386/i486 and such.
> 
> are this dinos still serving somewhere?
> 
> > Many people refer to i386 as all 32bit x86 systems.
> 
> I would say this preference is mostly set by beeing afraid of
> migration (lots of things can come up when migrating a production
> server) or by lack of money to buy some nasty HW ...

Ah, hardware bigotry.  Your colors are showing.

> > SATA (of any gen) still does not perform like SCSI.  Let's just look
> > at spindle speed alone ignoring the other benefits of SCSI.
> 
> I had no time to test it on a life webserver and probably can't do
> it so soon but I tell you that a 10K Raptor is faster then a 15K
> 320Mb SCSI when compiling world or untarring large files. Also NCQ
> is not reserved to SCSI anymore so when you see the price then it is
> becoming a valid option for small servers.

And your testing methodogy was...what?

Small servers?  No, let's talk about 'servers', not just 'small
servers'.

Very high disk I/O requires more than NCQ and 10K RPM disks, though if
you have a 'need' of disk space over performance, then SATA will be
your bitch as the cost (vs size) of SCSI vs SATA do change things.

Not all of us use small servers, though.

-- 
Mike Horwath, reachable via drechsau at Geeks.ORG