[CLUE-Talk] [Fwd: RFG-LinuxTCO-vFINAL-Jul2002]

[David Anselmi]

Jed S. Baer wrote:
[...]
> Yeah, and some of the Slashdot discussion was actually informative, like
> how they could have used a much smaller Sun machine than they did:
 >
>  * http://slashdot.org/comments.pl?sid=41797&cid=4411219

That's interesting.  He says this is a slam at Sun, but it may also 
represent the way Sun's customers do things.  You *could* run a web 
server on Solaris for less, but those surveyed aren't.  We need mo' 
betta studies (what else is new).

I'm puzzled at the way IT shops handle hardware.

In the study, they used samples of x86 servers from big name (expensive) 
vendors.  The hardware seems pretty commodity (except for SCSI, perhaps) 
  and they paid for same-day warranty coverage.  $5000/server.  I expect 
you can get the same for $2500 (even a few years ago).

So rather than pay $40k for 8 servers, couldn't you buy 10 for $25k and 
have ready spares on hand?  Is it really that costly for an admin to 
spend half a day swapping a failed server, vs. having an admin spend 
half a day calling for warranty service?

Anybody have any numbers (reliable study or extensive experience) to 
support this business practice?  Seems to be an outsourcing decision but 
in this, outsourcing is "best practice" whereas in IT generally there is 
some debate.

Dave