--
Office of the Provost                                   328 Tigert Hall
Information Technologies and Services        Gainesville, Florida 32611
                                                    Voice: 904/392-4519
                                                      Fax: 904/392-6886
                                       email: super@nervm.nerdc.ufl.edu

                  Oct. 25, 1994, Oct. 31, 1994

MEMORANDUM (originally via email)
To: Bob Cremer (for PPD), Mark Hale (for CIRCA),Tom Hintz (for
ICON), Ron Schoenau (for NERDC), Rick Smith (for Network
Services), Sharon Wheeler (for HEALTHNET).
From: Sam Trickey
Subject: Individual Response as called for in Recommendation 8,
Network '99

 In keeping with the process proposed on Oct. 18, this is my
response to the Recommendation 8 requirement for an independent
appraisal of
    a) the existing relationships among University networking
   organizations and personnel and their effects (pro and con)
 and
    b) the significance, merit, etc., of the so-called
   "integration concept."
Item a) Organizations and Relationships

 Seven organizational allegiances are represented in the people
charged with responding to Recommendation 8.  Some of them are a
lot more distinct than others.  The one common organizational
link (below the level that we all work for the SUS!) is
participation in the Council.

 The structure and processes associated with the Council have
done a lot of good in reducing tensions and misunderstandings and
increasing cooperation between/among these organization.
Certainly my understanding and ability to work with other
organizations has been helped.  More specifically, the
Distributed Management Protocol for Telecommunications, the Vice
Chair role on the NERDC Policy Board, and invited participation
in HSC's Information Technology Advisory Committee have been
useful steps.

 On a person-to-person basis the Council process also has helped
my functioning.  I believe that is true for others.  It seems to
me that we have come a long way in terms of respect and trust
after some very difficult discussions in the early going
particularly with HEALTHNET and PPD.

 Overall I think what we are doing is more-or-less getting the
job done (within funding constraints that are horrendous) in
spite of a rather awkward organizational structure.  For that
reason, there are still a lot of division-of-responsibility
issues that are unresolved or just sort of patched together.  A
historical perspective is useful.

 Data networking at UF evolved with several different overall
plans adopted by groups that largely didn't know about one
another.  Thus the NERDC SNA services came early and focused on
administrative users. NERDC involvement with FIRN attracted no
great interest in UF proper until recently, with the striking
exception of ICON.  Ethernet, tcp/ip, and Internet services were
driven by academics, originally in Engineering and in Liberal
Arts and Sciences, though NERDC Teleprocessing did participate in
the SURAnet effort right from the first implementation.  The
antecedents of HEALTHNET were pc LAN's and their interconnection;
the technical model and staffing both were in sufficient trouble
by early 1992 that HSC had to develop the current HEALTHNET
partnership with Shands in order to right matters.  CIRCA had its
own network architecture which derived from several influences,
VAX and pc architectures, staffing limitations, and several
remote labs among them.

 When UFnet evolved from the Engineering system to an embryonic
campus-wide system (driven largely by our 1986 SURAnet proposal -
that was where the name UFnet first saw light), appropriate
organizational and strategic change did not get made.  That is,
no campus-wide networking group with adequate funds and
management was formed. Instead, UFnet evolved piecemeal and
retained a sense of being experimental rather than production.
This evolutionary pattern became codified in the Campus Computer
Coordinators, a technical group that had to play a management
role because of the vacuum.  The management vacuum was aggravated
by several severe personality conflicts involving both networking
staff and various senior information technologies managers.

 The 1991 Task Force report recognized this same problem in
somewhat more cautious language; see the opening paragraph of
Sect. 3.3 and Recommendation 3.3.1.

 At the startup of OITS one of the things I discovered was that
there had been a quite normal reaction to the management vacuum
and the accompanying "experimental" emphasis: NERDC, CIRCA, and
PPD all were considering addressing the management vacuum by some
form or other of merger, incorporation, or absorption of UFnet
responsibilities, services, and positions.  What I have seen at
other universities and know from experience here didn't make me
happy with any of those choices.  This is not to knock the
competencies or integrity of any person or group.  It is simply
to deal with core strengths and weaknesses as I understand them.

 Specifically, the NERDC fee-for-service model doesn't fit well
with typical university networking funding models.  The Policy
Board has a long history of decisions which are perceived (I
think largely correctly from my experience in high-performance
computing) as inimical to academic computing (only since the
invention of the Research Computing Initiative in 1990 has that
perception been reversed).  Despite the "R" in its name, CIRCA
has little experience in the necessities of highly competitive
externally funded research. CIRCA also has little experience with
the nature of administrative networking.  PPD has virtually no
experience with large-scale data networking and the lessons
learned from telephony enterprises going into data networking at
the campus level are not encouraging.

 More important than any of these limitations, it seemed clear
that all three competing organizations had more than enough to do
in their own primary areas of responsibility.  Early on I
concluded that it was essential to begin to force the University
to come to grips with the need to fund a reasonable data
networking operation as a distinct organization in OITS.  (Put
another way, suppose UFnet were to be transferred to NERDC or PPD
on a fee-for-service basis.  Where would the new money come
from?) I also had to deal with the "management by techies" issue.
Therefore, during my second year I spent the largest single
fraction of my time on managing UFnet and on recruiting
management for it (a process begun unsuccessfully in year 1).

 Where are we today?  UFnet is a somewhat healthier organization
though still undernourished and small. We did manage to
consolidate the network wiring staff from Office of Academic
Support Services into Network Services.  UFnet reliability is
much higher than some old, hard to shed anecdotes.  I get very
positive comments on the architecture and implementation of the
three FDDI ring system when I show them to IT professionals from
other universities.

 Several of the technical-staff-level animosities have been
reduced or quieted, though the history of some of them is
sufficiently old that I suspect certain grudges will never be
allowed to be forgotten. We still have technical staff from
various organizations propagandizing against others, though that
too has been reduced dramatically.  I've tried to do my part on
these issues and am appreciative of others who have stepped up to
the task.

 Were it not for history, the next rational steps would be to
move some of the networking staff and support functions now at
NERDC and CIRCA into Network Services.  Since that is not
possible structurally with regard to NERDC and yet it is "our"
only clock-round production shop, we have contracted some of the
clock-around functions to NERDC. It must be admitted that the
contract is not a whole lot more than our making payment on what
UF was getting a free ride on earlier.  Ron Schoenau, Rick Smith,
and I have had some frank, candid, sometimes hard discussions and
email about further steps in that direction but have not yet
reached even a rough consensus on the division of
responsibilities.  We will keep working on it.

 I continue to believe, with the Task Force, that the directions
of telephony and data communications are such that
Telecommunications should, in principle, be a part of OITS.
However, I'm willing to see that concept tested in the external
review.  I also admit that the step would be a large and
difficult one and that the Distributed Management Protocol is
beginning to have some of the positive effects that a
reorganization would be intended to achieve.

 I do not agree with the off-expressed view that it is fine for
OITS to be involved with policy alone as long as some other group
gets to control operations.  All the policy in the world doesn't
mean anything if there are no mechanisms for making it take
effect in operations. The quickest way in the world to change a
policy de facto is to alter operations.  For that reason, I think
it is important, for example, that the Protocol with PPD is for
"management" not just "policy."

Item b) The Integration Concept

 When I drafted the original organizational simplification
recommendations for Council consideration (June 1992, on a plane
back from Germany!), I was a lot more concerned about some of
this integration concept than I am now.

 First off, there is a higher priority, namely the part of
Simplification Recommendation 2 that speaks to the need for a
Video Services Group (which we still don't have and need more
than ever) and the need to integrate telephony into Network
Services (see above).

 Second, OITS needs to get its own house in better order.
Because of the newness of Network Services and overall workload,
we do not have the relationship between CIRCA networking and
Network Services well defined nor have we ever considered whether
(and if so, how) to plan for their integration.

 Third, so long as ICON and HEALTHNET are operated (from the
point of view of UFnet) as extremely large subnets that are
responsible, have a compatible architecture, follow UF Preferred
Practices, etc., then their scale is such that continuation of
subnet-level management is probably in the best interests of
customer response, manageability, etc.  HEALTHNET of course, may
well be benefited in this regard if the HSC CIO position comes to
pass.

 Where I would draw the line on this subnet-level management is
the notion that such extremely large subnets are somehow peer
networks with UFnet, hence should be viewed as real or potential
alternative service providers.  UF doesn't have resources to do
the job once right, much the less be the sponsor of internal
laissez faire competition.