OT Linux (RE: [rescue] OT: Stuffed Proliant?)

[Gregory Leblanc]

On Fri, 2001-12-21 at 19:35, Patrick Giagnocavo wrote:
> On Fri, Dec 21, 2001 at 07:04:50PM -0800, Gregory Leblanc wrote:
> > Speaking strictly for Red Hat supported platforms (I.E. Not SPARC), the
> > VM in the latest kernels from Linus actuallystands up to their stress
> > testing without serious hacking.  I don't trust x86 hardware to
> > withstand heavy loads, so I tend to find that a pretty moot point.
> > 	Greg
> 
> That sounds a bit like flamebait actually.  IMHO it's the software not
> the hardware that makes x86 look so bad - since most of it ships with
> Microslop.

Wasn't meant to be flamebait, though I do have strong opinions on the
subject...  I honestly don't think that the chipset designers at either
Intel or AMD are good enough at their job to make something that will
stand up to seriously hitting the memory.  And the guys at Via and the
other third party companies, well, they're not even in the same ballpark
as Intel/AMD.  I haven't seen a chipset out of Intel that stood up to
heavy loads since the BX, though I heard rumors that one of their newer
ones was getting close to being that bulletproof.
Even if they did design the chipset to withstand serious abuse, most
motherboard manufacturers would skimp on the design, and use lower
quality/fewer capacitors to clean up the power that's getting fed to the
CPU/chipset, which could easily be enough to cause crashes under heavy
loads, when the whole system is getting stressed.  

> OpenBSD on x86 performs quite well in my experience.

OpenBSD is my firewall platform of choice, even though the Linux stuff
has stateful packet inspection.  I think OpenBSD 3.1 will have a bunch
of the features from Linux, and a -sane rules syntax-, which gives it a
pretty clear advantage.  I haven't tried it for any of my servers yet,
though when my New Toy gets here (thanks Ken!) I might play around some
more.
	Greg

-- 
Portland, Oregon, USA.