ULTRIX futures?

[bzs%bostonu.csnet at csnet-relay.arpa]

Doug Gwyn says that the future of UNIX is SYSV, and that is good.

The images in my crystal ball are not so clear:

Although several vendors have committed to some form of SYSV, we
see several major vendors (DEC,SUN, much of the workstation market
and near-supercomputing) showing development based upon 4.2bsd.

This is not an insignificant portion of the community. Although
IBM is toying with all possibilities there is a decided leaning towards
SYS/V (the PC/RT for example already has both SYS/V and 4.2bsd but
it looks like the SYS/V may be the only official release, in their
mainframe business almost exactly the same thing occurred but nothing
official has been released.) On the other hand, it is not at all clear
that IBM will ever be a significant factor in the *UNIX* market as
they continue to rabidly push MVS as the only serious O/S available
for their machines (I have spoken with people within IBM about this
and this message comes through loud and clear.) The IBM behemoth
is probably not a significant factor in this issue, at least not
for a few years (if you think what happens with PCs in general is
significant then I disagree fundamentally ['you' here is not Doug,
but you the reader, I believe Doug agrees with me on this.])

As we all should know by now, SUN has signed significant agreements
with AT&T to come to some sort of resolution of this issue, rumors are
that we should know what that means as early as the summer.  Obviously
SUN cannot throw out 4.2 (I guess obviously, at least they would have
to modify SYS/V to the point that it may as well be 4.2 to continue
their architecture.) I think AT&T is similarly not ready to throw out
SYS/V so the only good guess I can come up with is some compromise on
adding a lot of 4.2 function to SYS/V (and vice versa), a hybrid
offering, although I agree this is very optimistic.

Digital, as far as I know, is not likely to undo their current 4.2
based effort in ULTRIX. Most rumors to the contrary appear to actually
be referring to their acceptance of Doug's SYS/V emulation under
ULTRIX as a standard product, good for everyone involved. I believe
here this commitment is not insignificant to DEC, they must develop
a UNIX identity (as they should/could have years ago). Their VAX
line is destined for only a few more years (rumors from within the
company of a project referred to as 'The Last Vax' are already
circulating.) This leaves VMS entirely in the lurch (fantasies from
VMS fanatics aside of somehow porting the largely machine coded O/S.)
They can't make it cheap and they can't make it fast (if you think
otherwise you are looking at the world through vax-colored glasses,
an 8650 is not a good price-performer in today's market, neither is
a microvax II. I/O architecture on both machines may be their ultimate
downfall.)

This leaves DEC with only one choice: Release a new architecture
(rumors of this development effort have been con/persistant) and leave
VMS behind (remember TOPS-20...) This points towards a desparate need
to develop a solid image of their ULTRIX product which will lead them
into this new market (with 'transition aids', note the offering already
of VMS/Fortran under ULTRIX, these popular and profitable applications
will be a lot easier to port than the underlying VMS O/S.) I doubt
very much DEC is about to undo what they have so far established at
this critical junction in their history (the company without a product?)

The underlying implication in this whole argument is that somehow
AT&T, by virtue of the fact that they own the concept of UNIX, is
pre-destined to become the industry leader in delivering UNIX systems
to the world. Oweing to this leadership position they will thereby
succeed in proliferating their SVID as a standard. Obviously their
raw perceived size as a corporation is also a large factor in this
prediction.

I for one disagree. We are one of the gift sites for 3B2 and 3B5
systems and have been running these for over a year. I have been
in contact with many other gift sites and several other customers.

There seems to be a lot of agreement (not unanimous, but nearly
so) that ATTIS in fact is one of the most mediocre vendors you
can get involved with. Their systems are unreliable and unexciting
being mostly slow and very limited in function. The organization
itself is nearly impossible to deal with, even when it's in their
own best interest (eg. calls to try to find out about upgrades get
unanswered, field service people appear untrained and of an attitude
that putting a system down a few days to work out a problem is fine.)
Worse, their product pricing is confusing at best, in fact many consider
it abusive (unbundling pieces like t/nroff with each new release,
people find they have received systems wherein the standard release
lacked even a C compiler.) They act as if you have no choice (a
situation with a familiar 'ring' to it...We're the phone company,
we don't care, we don't have to, we're the only game in town...)
My very typical story is I am in the 12th month and waiting for
delivery of pieces of my system, things do not get resolved.

Further, promises of some proliferation of applications has failed
to develop. In the field of networking, for example, most people I
have spoken to find 3Bnet 'a joke', not to be taken seriously (it
basically consists of one command, no application interface, imagine
a subset of uucp over ethernet and you are close.) There are promises
of an 'RFS' but again one cannot help but notice that any homogeneous
networking solution (ie. 3B only) is not destined to be very popular
(remember all those Vaxes and IBM mainframes out there, they're not
going away overnight, not to mention workstations.)

If we remember other corporate giants who decided to enter computing
as a business (General Electric, Exxon, Xerox/XDS, RCA, Raytheon)
we see a clear pattern of initial assumptions that of course such
a large, well backed adventure will make an impact, in fact all of
these ventures failed. There have been exceptions to the contrary
in recent history (of course IBM and Burroughs might be used, but
their entries were so early that they virtually monopolized their
areas quickly) but they are few and far between (name 3.)

It is not at all clear to me, at this point in history, that ATTIS
will not in fact start to realize how difficult it is to actually
run a business in computing, coming out with products, service and
support in a timely fashion. Their massive bureacracy seems to hinder
them by making products which people have never even heard of come out
badly supported, expensive and with no clear identity in the marketplace
(WHY would YOU buy a 3Bx given all the other choices on the market? If
it's only for recursive reasons [it's from AT&T] you are destined to
a hard lesson about product assessment.)

I think it very likely that they will fall back into their traditional
role in telecommunications (PBXs etc) where they find acceptance and
know how to compete and leave the very fast paced, competitive market
of small to medium sized computing systems to the lean and mean startups
and the very few corporate giants with established bases to fall back
on.

This would mean a waning in interest from ATTIS and hence dominance
by the current giants in the UNIX market (DEC, SUN et al.)

I guess what I am trying to say is: The emperor may very well have
no clothes.

As I have said over and over again, it's still a lot better than
anything else...except 4.2...

	-Barry Shein, Boston University