[rescue] intel vs. sun- for real

[Joshua D Boyd]

On Fri, Jul 19, 2002 at 08:50:55AM -0700, Koyote wrote:
> So, my girlfriend is being forced to give hard evidence why a single
> or dual cpu UltraSparc would be more advantageuous to purchase than an
> intel box. Performance, reliability, anything. 
> 
> Of course, background is needed: this is mostly a NFS, HTTPS, and
> samba server.(There's a little bit of website front end, but it's 
> mostly about getting data to user.) It's internal, serves large
> amounts of data per connection, but sustains relatively few
> connections. - though the mean load is ridiculoudly low (call it 20
> downloads per day (except at beginning of term), but 50-700 meg each
> download) 

So, on average you would be transfering 14gigs per day?  But what is
the peak.  Or is it ok to force people to wait until the load is
lower?  

Because so far with the average load you give us, you make it almost
sound like a lowly U1E, or whatever the cheapest machine with extra
expansion slots that is still supported, would be able to the job just
fine with an extra SCSI card and extra fastE card.  If this is the
case, why not buy such a system, and test it out.  If you can prove
that a U1 or U30 would do the job pretty well, then it should be a
piece of cake to get the cheapest equiv. machine sun still sells and
outfit it with disks. 

Well, perhaps I'm being idealistic.  But it just seems to me that this
is a case of needing to actually get and test both machines.
Hopefully you are an institution that would be able to get them as
loaners as opposed to needing to buy them, in which case your battle
is lost most likely.
 
> So, what is needed is hard, real, hard numbers. Which I'm having a
> hard time finding. Redhat vs. Solaris. US2 vs P4, Sun performance vs
> Dell performance.... 
> 
> Any real data out there?

The only true real data for something like this will be your own tests.

-- 
Joshua D. Boyd