ufl or uf

Executive Summary

In November of 2007, I solicited comments from the CCC (Campus Computer Coordinators) on an issue which had been discussed before the November ITAC-NI meeting; "A request has been made to update the domain from UFL.EDU to UF.EDU" [1] [2].

Comments submitted to the list list make it apparent that many members of the University's information technology community are not in support of changing UF's domain from "UFL" to "UF."

Responses to the suggestion were complex and expressed a broad variety of concerns. This document attempts to distill the many points, some repeated, into a concise summary.

Basis for the following responses

We found it difficult to interpret the scope of the desired outcome.

The requirement may be simply that www.uf.edu resolve to our home page, and that mail addressed to 'joedoe@uf.edu' finally be delivered to the GatorLink user 'joedoe'. Half of this work is already done.

However, the requirement statement implies that the ufl.edu namespace is to be deprecated immediately if the project is approved, and expunged as quickly as practical thereafter. Satisfying this interpretation would be exceedingly complex, and has communications and reputation costs to the University and its faculty. It is on this interpretation that these comments have been based.

Questions and concerns

General

External impact

Maintenance of "legacy" namespace

It is not clear how long the legacy namespace would be maintained. This decision would have impact internally and externally, confusing our points of presence to the world, and also to our own community. Some data points on the question include:

There are tensions in both the "short" and the "long" directions for this decision. We want the 'ufl' around as briefly as is possible because the dual namespace will dramatically dilute our search rank. On the other hand, we want to maintain the clear communications channel as long as we can: we do not want to turn aside communications simply because we pulled up stakes.

Damage to established presences:
Damage to google page rank

Since any page would be reachable by two independent addresses, current page ranks would be sharply curtailed as some audiences learned of the revision, in essence splitting our reputation between the two sources. Once we deactivate the legacy namespace, there will be a population of dead links which will not benefit our SEO (Search Engine Optimization) at all.

This consideration encourages short legacy maintenance; we want to make old links fail as soon as possible, in the hopes that someone will notice the failure.

Damage to image: perception of priorities

UF is in the middle of a hiring freeze, and we have seen harbingers of serious budget cuts. In that environment, spending large amounts of time and effort on cosmetic modifications should be very carefully supported with an appreciation of the hard benefits which will devolve from the initiative.

Appearance of spam: failed reverse lookups to legacy namespace

When we deprecate ufl, reverse lookups will return .uf domain names. This means that any mail proceeding from untransitioned .ufl addresses will appear more spam-like because reverse DNS for it will fail.

Internal impact

The internal direct costs of implementing the desired changes are real but complex. No unit can make responsible estimates without thoughtful and extensive evaluation, work on the order of that required to actually implement the change. What we can do, however, is establish a baseline list of tasks which we know any unit must accomplish.

Having produced this, we could multiply that minimum effort by the total count of units having net presence, but even this extrapolation is problematic. The lines between "units" are sufficiently changeable, subtle and ad-hoc that Institutional Research doesn't attempt to calculate them.

That said, here is a minimum list of tasks which will need to be performed in every UF unit in order to support the suggested change.

Modification of existing documents

Identification and audit of custom code

There are thousands of pieces of custom code which have been written in the last twenty years which reflect the presumption that the domain name is stable. Examples include, but are clearly not limited to:

Every department would need to audit all of its functions and supporting software to identify problems.

Licensing issues

Many software packages are licensed based on domain boundaries. The contemplated change would require revision (and perhaps renegotiation) of every license so expressed.

Turnover in namespace enabled by forcing a change

Since we will be directing that every department change its name in any case, many of them will opt to change to something other than [their-old-third-level].uf.edu. This will result in a complete re-fighting of the third-level domain battles.

This will generate substantial internal conflict, and profoundly amplify the external and internal confusion because it will not be possible for anyone to simply replace 'ufl.edu' with 'uf.edu' in their URLs.

Similarly, within a given web presence, the renaming will likely generate the incentive for re-evaluation. Though a department might well decide it is time for a revamp, changing their web structure at the same time we undergo a top-level change will exacerbate the confusion in mapping between namespaces.

AD modifications
It is suggested that modifications to the AD root domain are extremely difficult [technet2.microsoft.com]. These changes would have pervasive impact on AD-connected departments.

Conclusion

We believe that before a decision of this magnitude is made a true ROI (return on investment) analysis should be done. This project would need to include university PR, IT, and Accounting staff to make a proper evaluation.

As it stands, we feel that the proposed change has not been specified with sufficient clarity to allow a proper analysis. However, we are concerned that the costs in UF reputation, direct costs, and time-worked could overwhelm the benefits in making this proposed change.

Our group will be happy to answer any IT questions you may have related to the proposed domain change and would be equally available to assist in a business ROI analysis.

Some Links and references.

[1] The official project plan for the transition

[2] The Minutes of the ITAC/NI meeting at which the subject was discussed in detail.

Most of the opinions here distilled were expressed in the CCC discussion on the topic in November of 2007. References: