Re: Dell stops offering Red Hat on desktops

[Sean /The RIMBoy/ <sean@rimboy.com>]

On Tue, 7 Aug 2001, Steven Pritchard wrote:

> > Our actual operational uptime for a cadre of over 1100 servers in ITG with
> > a wide mixture of workloads for a recent six-week period was 99.91%, and
> > almost none of these systems were clustered. Clearly Windows NT is highly
> > reliable when operated competently on good quality hardware and properly
> > maintained with hot fixes and service packs.
> 
> There are Linux clusters ten times that size.  I wonder what sort of
> average uptime they get...  Probably better than an hour of downtime
> in a 6-week period.  (Do the math.  It's actually pretty pitiful.)

Actually, the largest known Linux cluster is NCSA's Platinum, checking in
at 512 nodes (1024 procs).  That said, 1024 is the next mark to break, and
several places, including NCSA are working to break that mark.

That said, NCSA's Platinum are IBM's Netfinity servers (1U
systems).  According to one of the guys, during the shake down they were
loosing at least a stick of ram a day during testing.  And of course, he
said, it was always the 1 gig stick, never the 512 meg stick.  

That said, Google apparently has a least 1-2 systems tank a day, but since
they all apparently contain a small chunk of data, it is transparent to
the user.  

Yes, the failure rate drastically increases when the number of nodes
increase, but then again, it's all stats.  

Sean...

-- 
WWJD?  JWRTFM.     -ASO/.
_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_
KG4NRC
www.rimboy.com		<-- Your source for the crap you know you need.

-
To unsubscribe, send email to majordomo@silug.org with
"unsubscribe silug-discuss" in the body.