Alfredo says [among other wonderful things that helps put these
issues into perspective]:
>
> Just in case some gentle HP3000-L participants might think that there is a
> Unix vs. MPE "war" going on, I would like to remind them of a couple of
> facts:
>
> Tough questions, indeed!
>
 
As a contributor to the Unix vs MPE "war", perhaps I can clarify some of the
issues we have.  Some background is important, so this will be a bit long.
If your not too interested, just skip to the end (or delete and move on).
 
Last year, we were in the market for a new, medium size, computer server.
We have been a long time MPE customer (series III) and we value our
MPE systems very much (currently on a 967). We have used the 3000 for nearly
anything "computing" related in our university enviornment, from CAI to
programming instruction [remember Robelle's pascal?] in Computer Science, but
mainly for administrative systems (Image and Powerhouse w/a bit of Cobol).
As a result of this experience and our ability to keep staff turnover low, we
have staff well skilled in MPE.
 
Along came the Internet and the multi-media IT "revolution." We were
fortunate to have a sizable hp9000 workstation (a 730) and, with HP's help, we
established internet email accounts for faculty, setup a DNS, and
generally got into things Unix and Internet. We also setup a new department,
the Center for the Advancement of Teaching, to promote faculty development of
multi-media information technologies.
 
It soon became clear we would need a larger machine if we intended to
provide full support for WWW usage in classes, email for all students, etc.
We were again fortunate to have the funds to purchase a larger (medium size,
really) server, but these funds had to be spent "now" or they would disappear.
 
Since we had the long relationship with HP and MPE, our initial reaction was
to go that way.  The 9x9/Kxxx series were just announced and looked like they
would take us far into the future (processor upgrade support, high bandwidth
processor bus, etc.).  We looked at MPE, we looked at HPUX (Web, Email,
Groupware, etc.).  With Unix, we were concerned about the training costs to
develop serious HPUX skills and about system reliability and security
(compared to MPE), but software availability was critical.
 
Bear with me as we look at each of these. For the Web, httpd was unsupported
on everything, but obviously it would be sure to be updated and maintained for
HPUX - not true for MPE as HP wouldn't commit to it. This is a technology we
would have to learn, no hope for leveraging what we knew.
 
For Email, OPenMail was obviously the HP solution - even though we were long
time users of HPDesk (like HP itself), HPDesk looked dead in the water. Even
so, I rated HPDesk as a solution equal to OpenMail (depending on the long
term support I had seen HP provide for other MPE technologies). On the
training issue, HPDesk won hands down - we would have to invest in a some
batch type account maintenance, but the facilities for this were available and
supported.
 
For Groupware, we tentatively selected Lotus Notes.  Notes was just announced
on HPUX and there looked to be no hope for MPE.  We have since put off a
Notes implementation, feeling that what we need Email or the Web takes care
of and to costs a lot, plus it is not really scaleable (using a single system
solution).  Again, we expected to have to learn this technology and the
serious cost of this also contributed to tabling it.
 
We turned to the people we knew at HP and almost every person we contacted
recommended HPUX and the new K series box, including our current and former
MPE ASEs and (of course) our sales rep.  On our concern for the cost of
training and developing Unix skills, they responded that this was an
investment in the future, but moreover, HPUX has superior "easy of
use/admin/operation" etc.  products under OpenView and with SAM.  By
investing in these products, we would save on training costs and be more
effective more quickly.  On our concern for system reliability and security,
HP assured us the with HPUX 10, HP was begining to deliver MPE type
commercial reliability and security on Unix.
 
None the less, we priced both solutions, with the recommendations for the
OpenView products (not OpCenter) and training for Unix.  This was before HP
announced the "small" user versions of the 9x9, so we had to look at a 64 (I
think) user version.  The MPE solution was competitive, but we obviously were
not buying nearly as many new products or training.  This, with the serious
doubts about software for MPE, and HP's assurances about HPUX from HP's
representatives, we selected the HPUX/OpenView/training approach on a
K400/200.
 
Unix, we knew, would provide a "latest and greatest" environment for whatever
new things appeared on the Internet, but we could provide this by maintaining
our existing HPUX system. What we were in the market for was a commercial
quality Web and Email solution for the entire campus.
 
Since that purchase, MPE now has a commercial Web server solution (thanks to
OpenMarket) and OpenDesk has the "almost right" Windows GUI with a "classic"
secure future (as I expected), but still no sendmail and no DNS, but telnet
inbound is really coming.  It would be a *lot* cheaper now to go with MPE,
given the smaller user licenses now available for MPE.  (I need to note that
we don't need many sessions, since the Web and Openmail both are
client/server).  Even so, CSY has not shown (to me at least) a clear
commitment to the tools and services necessary to make MPE a commercial
solution for the Internet/WWW. The speed at which the industry moves has
increased - witness Microsoft's reorganization - CSY needs to find a way
to respond more quickly without compromising traditional MPE system integrity
and protecting our investment in MPE. Commercial quality computing now
includes Internet/WWW services and not just OLTP.
 
The K system itself is too new to really evaluate.  So far, OpenView
(OmniBack, OpenSpool) and SAM appear to be a really nice, easy to use,
solution to operations and administration - MPE could use these
(although some steps have been taken with the on-line configuration tools)!
HPUX 10, however, has too many holes (like choosing between JFS and ACLs!),
and the late arrival of many HP software products.  The staff still have a
long way to go on Unix skills, and I wonder what will be the cost to our MPE
skills. HPUX 9.x system reliability on our 730 has been ok, but I get tired of
killing run away processes and sessions that no longer have anyone connected
to them - at least it is easy to reboot from home.
 
Today, I don't know if we would have made any different decision.  Neither
approach is really satisfactory.  What I really wanted was an MPE system,
priced for our need, with a commercial quality set of Internet/WWW software
services and support we needed, and HP's personnel to "believe in" the
solution.  Much could also be said about software tools suppilers like
Cognos, that have yet to come up with any support for linking Powerhouse with
the Web.  (Yes, of course, we can learn to do it, but we bought a 4GL so we
wouldn't have to code at that level -- besides, we are busy learning to run
an HPUX box and to design Web pages.) As it is, now I contemplate life without
MPE when I consider and plan for possible futures - never did I do that before!
 
Hope this helps!
Richard
--
>>> - Speaking for myself and not necessarily anybody else - - <<< NOTE!
Richard Gambrell        | Internet: [log in to unmask]
Mgr. Tech. Services     | POT:      504-483-7454     FAX: 504-482-1561
Xavier University of LA | Smail:    7325 Palmetto, New Orleans, LA 70125