SUMMARY: Solaris: x86 versus Sparc?

From: Erwin Fritz (efritz@GLJA.com)
Date: Tue May 25 1999 - 13:36:24 CDT


I received many, many replies to my original query, which is posted below. By
far, the clear winner was the Sparc platform.

Choosing the Intel platform to run Solaris had only one advantage over the Sparc
platform: a cheaper initial purchase price. Note the word "initial", since
subsequent maintenance and administrative costs would make it higher than
Solaris.

Here are the advantages of the Sparc platform, in point form:

- better hardware/driver support
- faster bus architecture
- larger on-chip caches (4 and 8MB)
- 3 floating point processors in each CPU
- better availability, because you can remove system boards while the E4500 is
up and running
- scalability
- existing binaries from my Sparc 1000E will run on the E4500 without needing
recompiling
- one service contract will cover both the OS and the hardware
- 64 bit hardware and a 64 bit operating system (Solaris 7)

Thanks go to the following (assuming I haven't missed any)

 Al Hopper <al@logical-approach.com>
 Bertrand Hutin <hb@ardentsoftware.fr>
 Bill Hebert <bhebert@nuc.berkeley.edu>
 Birger Wathne <Birger.Wathne@Ark.no>
 Brian Friday <bfriday@LaSierra.edu>
 Brion Leary <brion@dia.state.ma.us>
 Brooke King \(6532\) <jbking@sandia.gov>
 Carlo Cosolo <ccosolo@tse.com>
 Cathy L Smith <csmith4@fwhns41.ftw.mot.com>
 Chad Price <cprice@molbio.unmc.edu>
 Charlie Mengler <charliem@anchorchips.com>
 Choi, Eugene <choi@ftintl.com>
 Chris Eslinger <eslingc@atd.sprintcorp.com>
 Damon LaCaille <Damon_LaCaille@dgii.com>
 Dave <dburwell@symmetricom.com>
 David Evans <DJEVANS@au.oracle.com>
 Douglas Palmer <palmer@nyed.uscourts.gov>
 E Muller <emr@wcomp.gov.za>
 Franczyk, Gary <gfranczyk@carbomedics.com>
 Gary Smith <gsmith1@ftw.mot.com>
 Greg Obremski <obremski@fdu.edu>
 Henry Pan <henry@public3.bta.net.cn>
 Jim Bacon <jim@nortx.com>
 John Dorsey <dorsey@Colquitt.Org>
 John Weekley <weekleyj@inlink.com>
 Justin.Stringfellow@barclays.co.uk
 Kai O'Yang <oyang@earth.pscit.monash.edu.au>
 Matthew Fansher <fanshem@gcm.com>
 Matthew J. Rock <mrock@intel-sol.com>
 Pat Hooper <phooper@nebs.com>
 Peter L. Wargo <plw@ncgr.org>
 Petri Kallberg - Sun Finland - <Petri.Kallberg@finland.sun.com>
 Randy J. Parker <randy@mobiledyne.com>
 Rick Robino <rrobino@wavedivision.com>
 Ronald Loftin <reloftin@syr.edu>
 Scott F. Woods <sfw@adc.idt.com>
 Stephen N. Carter <carter@accj.or.jp>
 Steve Harris <harris@dante.gsfc.nasa.gov>
 Studebaker, Mark R <mark_studebaker@reyrey.com>
 Thor Newman <newmant@starnetc.com>
 Timothy Lorenc <lorenct@load.com>
 Varad Rajan <varad.rajan@hsc.hac.com>
 Vince Merrell <blipvert@blipvert.screaming.net>
 dana@dtn.com
 ian@masuma.com
 kevin@joltin.com (Kevin Sheehan {Consulting Poster Child})

-------- Original Message --------
From: Erwin Fritz <efritz@GLJA.com>
Subject: Solaris: x86 versus Sparc?
To: Sun Managers <sun-managers@sunmanagers.ececs.uc.edu>

We are looking at replacing an old Sparc 1000E with a new server, probably an
E4500. The machine would be an application server for the users, most of whom
telnet it to it.

A sales rep I deal with (but not for Sun stuff) claims that he can put in a
Compaq (Intel-based) box instead, and that we could run Solaris x86 on it. He
claims that this should be as fast as running Solaris/Sparc on the E4500. Of
course he'd say that; he makes the sale that way.

Now, I've only run Solaris on Sparc architectures. Also, I don't have any
real-world experience with the E4500. I'd hate to go with his recommendation,
and then find out that Solaris on x86 is as troublesome as NT is.

What are your opinions on this? For a mission-critical system (which this is),
is it better to stick with Sparc CPUs when running Solaris? Is that faster?

Does anybody out there run Solaris on both Intel and Sparc processors? If so,
what's your opinion?

-- 
Erwin Fritz
UNIX/NT/LAN/DBA Guy
Gilbert Laustsen Jung Associates Ltd.
http://www.glja.com



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:13:20 CDT