[rescue] Fry's

[Dave McGuire]

On February 22, Rick Hamell wrote:
> 	If you want to talk speed, well, I don't know of any Sun Machines
> around that do 1.3 terrabytes of data transfers like ftp.cdrom.com use to
> do (up until the last transfer of FreeBSD ownership.) 

  Uhh, what?  1.3TB/day isn't all that much.  There are fairly old Sun
systems (and other real computers) all over the place doing much more
than that.

> 	Feel free to flame me, I'm willing to learn from people who have
> more experience or different experiences from me. :) But, I will not ever
> believe that one architecture is "best".	

  And well you shouldn't, of course.  But there are usually best and
worst architectures *for a given task and workload*, and PeeCee
systems are rarely (if ever) the best for ANY task...despite of sad
attempts by the suited world to press them into service where they
don't belong.

  Speaking in terms of system architecture, the PeeCee was such a bad
design to start with that certain attributes are "locked in" by
design...and any new system that needs to be PeeCee compatible must
implement those bad design decisions, because to change them would
render it incompatible.  The lack of foresight (and, quite simply, ANY
engineering skills) on the part of the IBM engineers that designed the
PeeCee is by far the most impressive thing I've ever seen.  Only in
very recent years have designers found ways to build PeeCee-compatible
systems that avoid some of the more egregious limitations in the
original system architecture...only now, when the Intel x86 processor
line has almost completely run out of steam.

  If you'd front-end a Sun database server with a PeeCee, I'd suggest
looking a little more closely at those performance and reliability
figures...not to mention little things like manageability.

    -Dave

-- 
Dave McGuire
St. Petersburg, FL         "Less talk.  More synthohol." --Lt. Worf