Subject: Re: Community Issues ** LONG **
To: Ken Nakata <kenn@synap.ne.jp>
From: Erik E. Fair <fair@clock.org>
List: current-users
Date: 02/22/1999 11:44:34
Sun approached the FreeBSD people about doing an UltraSPARC port that Sun
would pay for. When Jordan pointed Sun at us since they were already
committed to doing an Alpha port and had no additional resources, Sun said,
"No, we want *FreeBSD*."

I took the following impressions from this:

1. It was Sun marketing, not any technical people.

2. They noted that FreeBSD on Intel boxes has been making big inroads in
the ISP server market, at Sun's expense.

3. They want to compete in this market, and since FreeBSD has name
recognition, they wanted to be able to say that "FreeBSD runs on our
hardware too!" (I'm sure they're aware of the SunOS partisans who are still
pissed off at Sun about Slowlaris).

4. I doubt that Sun is sufficiently committed to this (or any) market to
match the basic price/performance of commodity PC hardware with the
appropriate price cuts on Sun hardware, so the effort to get FreeBSD on
UltraSPARC for this purpose was probably wasted anyway.


If we want to play in this arena, we need not only to package our software
a little better than we do (and include things in the basic package that an
ISP would want - think "solutions marketing"), we need to better document
not only what hardware we run on and what peripherals we have drivers for,
we need to document which hardware performs the best, which peripherals are
better designed and/or less trouble-prone, and where to buy!

In short, we need people for this who are committed to doing it and
maintaining it, and we need support from the code hackers and driver
writers who can say "this hardware is great; the driver was easy to write
and it performs to the limits of the interfaces" or "this hardware sucks,
it was designed by a brain-damaged mental patient, and no one in his right
mind would buy one."

This information is absolutely required by anyone who wants to run NetBSD,
but has no idea what to buy. They need to know what works, what doesn't,
and where they can get it.

Ideas? I got some. How about starting with the obsolete equipment vendors
who handle things like old Suns, HP's, 68k-Macs, etc. Particularly for the
really old gear, wouldn't it be nice for them to know that a modern UNIX
can run on that old stuff? They might be able to move out some of their
dusty inventory...

	Erik <fair@clock.org>