*BSD News Article 4954


Return to BSD News archive

Newsgroups: comp.unix.bsd
Path: sserve!manuel!munnari.oz.au!uunet!usc!sdd.hp.com!network.ucsd.edu!qualcom.qualcomm.com!servo.qualcomm.com!karn
From: karn@servo.qualcomm.com (Phil Karn)
Subject: Re: [386BSD]
Message-ID: <1992Sep13.060959.26409@qualcomm.com>
Sender: news@qualcomm.com
Nntp-Posting-Host: servo.qualcomm.com
Organization: Qualcomm, Inc
References: <p5tullo@sgi.sgi.com> <Btr5F5.AJ9@pix.com>
Date: Sun, 13 Sep 1992 06:09:59 GMT
Lines: 64

In article <Btr5F5.AJ9@pix.com> stripes@pix.com (Josh Osborne) writes:
>But I don't think 386BSD _currently_ has support for PPP, or AX.25 (and
>there may be a few other things 386BSD doesn't have), also NOS seems to
>lose way fewer charactors at high speeds.  Hopefully we can fix all that,
>but for the moment "NOS is not needed" isn't quite true...

I may be somewhat biased, but I agree. :-) I am at this moment using
NOS as a demand-dialed IP router between my home Ethernet and a
Netblazer at work. It supports a 386BSD machine and a DOS machine on
the Ethernet side and a Codex 3260 V.fast (24.0 Kb/s) modem on the
serial line side.  It is really quite nice to have a separate,
dedicated machine doing the SLIP routing, especially while I'm still
struggling with my BSD stability problems. Those old stripped 286s
that aren't good for much of anything else these days make great
dedicated routers, especially for SLIP lines where they can offload
all those character interrupts. (Remember when we all used to use
Vaxes and Suns running 4.2BSD as IP routers in our company
networks? That didn't keep Cisco, Wellfleet and Proteon from
building successful businesses.)

I began NOS in late 1985 to bring multitasking TCP/IP and its basic
applications to hardware that the average ham (or student) could
afford. (It actually began on a dare, when Terry Fox, WB4JFI, a rabid
CCITT/X.25 supporter, insisted that TCP was so monstrous that it
wasn't possible to implement it on anything less than a big VAX. I've
since discovered that the best way to get me to do something is to
insist that it is impossible.)

Given MS-DOS's popularity (like it or not) I've succeeded -
technically. But there are still many hams who are daunted by NOS's
(and TCP/IP's) relative complexity. So it's still a minority player in
the amateur packet world, which is still dominated by the "dumb
terminal and BBS" model that was state of the art in, oh, 1979 or so.
C'est la vie. You can lead a horse to water, and all that.

Now given that a typical UNIX system is several orders of magnitude
more complex than NOS and TCP/IP, I fear it'll be a long time before
the average ham will have his own 386BSD system (or whatever succeeds
it.)  The cost of the software or even the hardware is no longer the
limiting factor. It's now the user's ability to deal with what he
perceives as overwhelming complexity.  Face it, those of us who really
know how to configure, operate and maintain a networked BSD UNIX
system are still rare enough to be in high demand even in a recession.

But I do hope that there will eventually be at least one ham capable
and willing to provide a UNIX or 386BSD server system in most ham
packet communities. The rest can use NOS to access it, once they
figure it out, of course.

Some people do seem to be trying to turn NOS into another UNIX
lookalike this by (re)implementing large and complex applications,
many of which are already well supported in BSD, like netnews,
sendmail and named.  That was never my intention.  Now that 386BSD is
finally out, the fancy applications can go where they belong and NOS
can do what it does best: supporting dedicated or specialized
applications like routing and AX.25, and serving as a relatively
low-overhead platform for protocol experimentation. I know that for
amateur packet radio at least, 386BSD and NOS are a highly
complementary pair. I'm very excited about the possibilities,
but the hardware/software costs and technical problems are the
least of our worries.

Phil