|| EVENTS |
|| THE 17TH LISA CONFERENCE |
October 28-30, 2003, San Diego, California
USENIX home page
LISA'03 home page
Solaris x86 Community BoF page
Solaris x86 Community BoF
by John D. Groenveld
- Eric Boutilier and Alan DuBoff spoke for Sun.
- Phil Brown spoke as Sun's guest from the community.
Sitting next to me, taking notes was Cindy Swearingen who was great
to meet face-to-face as she's been on receiving end of previous
Eric was to give a short marketing presentation on where Solaris x86
is today and going to be in the future. I think it stretched a bit
because the crowd became eager to give feedback about the slides.
Sun gave out two cases of S9 x86 2003-08 media and most of the crowd
stayed until the end despite the small room and limited number of $BEVERAGE.
Phil Brown gave a short presentation on FreeHA, Free High Availability
At the end Alan offered to help folks install Solaris on their laptops.
Phil and I helped David Gawley from UWaterloo setup his Compaq C600
with an embedded Intel 10/100 which is missing from Solaris' devicedb.
Alan also spoke about Sun's interest in getting more entries into
/etc/driver_aliases and /boot/solaris/devicedb/master.
One major issue is that its Sun's preference to have these devices on the
shelf somewhere inside Sun so they can be locally certified and supported.
Alan mentioned that he's working on a document for BigAdmin to assist users for
the time being.
I didn't hand out the Solaris case badges as promised because they
hadn't arrived before leaving for San Diego. No one complained, but
hopefully someone going to SunNetwork Berlin's Solaris x86 session will
volunteer to hand them out. Or maybe I'll just go.
Some of the feedback Sun received from the community was rehash of
stuff discussed in the SolarisOnIntel mailing list:
The free RTU explicitly exempts dual CPU systems though those are
becoming more prominent. I recall the attendee pointing out that
a dual Athlon desktop had great price/performance.
I asked about B100X/B200X's estimated ship date and someone else
in the crowd said that he had heard January.
A couple folks shouted out about SunRay Server for Solaris x86. Apparently
they had applications where having x86 in the backend would be a win.
After seeing Eric's slide about Sun Solaris vs RedHat AS,
SuSE (Novell-IBM) ES, Microsoft BS TCO, there was a fairly long
winded comment from someone in the crowd about how great Solaris
is both on pricing and on support life and upgrade path.
There was much head nodding.
Eric asked about moving lxrun functionality into the kernel .
It did not excite the crowd. Presumably because most folks just want
more Solaris applications. Alan mentioned OS/2's Win/OS2 conundrum.
Eric talked about Solaris Express and Sun's release train. Someone
complained that managing quarterly releases of given Solaris versions
had some issues and asked if there was something other then
'cat /etc/release' to determine which release was installed on a system.
I suggested maybe a uname(1M) option. Phil mentioned that as far as
software dependencies go, ISVs and customers should explicitly look at
patch revs and package versions.
There was a comment about x86 vs SPARC performance. There was
agreement among the crowd that cycles per second was less important
than real world, customer work completed per second.
Cadence was mentioned as an example where x86 beat SPARC. Someone
asked Eric if Cadence was being sought after as a Solaris x86 ISV
and he mentioned that a whole bunch of vendors were being pursued.
Someone else from the crowd said that Cadence competitor, Mentor
Graphics, was considering a Solaris x86 build and that customers
should inquire with Mentor's sales office.
Eric showed a slide on Sun's new lower support pricing which I think
included numbers for both "Software-Only" and support bundles for
Sun's x86-based systems. I complained that it was like jumping
through flaming hoops to get "Software-Only" support contracts and
Davide Gaetano from Georgia State suggested I seek out my regional
Sun support sales rep and also apply more pressure on my systems
Someone, maybe me, asked about Fortran, Grid, various HPC components
ETAs. Eric replied that these were on Sun's ToDo list. Jed Dobson
from Dartmouth volunteered to introduce me to his contact in Sun
who eats, drinks, sleeps HPC.
Eric mentioned the recent IHV announcements regarding FC HBAs.
Joe Rhett complained that ASUS and other IHVs which used to include
Solaris Ready logos on their boxes no longer do so and he hears that
they're ready and eager to so again once Sun asks them. Alan assured
him that Sun's working them as fast as possible.
Eric said the slides from his presentation, particularly the newest
Sun Solaris vs the world TCO slides, would eventually end up on sun.com.
I think he was referring to something like