ITAC Minutes 21-Dec-95
Submitted by Anne Beckwith
In attendance: Robert Wolpert, Mike Pickett, Betty LeCompagnon, Anne Beckwith, John Board, John Brehm, Annette Foster, Roger Loyd, Melissa Mills, Steven Natelborg, Caroline Nisbet, Richard Palmer, Dietolf Ramm, Charlie Register, Marion Shepard, Tom Sibert
Robert Wolpert called the meeting to order.
In David Jamieson-Drake's absence, Mike Pickett reported on what he knew about the current status of the Data Subcommittee based in part on his recent conversation with Suzanne Maupin, a member thereof. Suzanne has gathered information about and reviewed the state of several other institutions' data models and concluded that the current state of sophistication of global data models was not extremely advanced. She will share what she has been able to obtain with the committee. Mike has been approached by vendors eager to work out attractive deals for Duke to purchase their database server software as a data warehouse platform and has arranged for them to meet with the committee; caution was urged before committing to a specific vendor / platform due to the tendency to be "locked in" to that vendor's products once the initial investment is made. Mike discussed a powerful new decision support / reporting application called Brio which he had recently seen demonstrated. Along with its ability to make multiple SQL table joins transparent to the user, it can rotate rows of data into columns so that analysis can be performed through records as well as across columns in tables. Someone (John Board?) pointed out that the recently discontinued Lotus Improv software had this capability, but had not been a big seller. Mike Pickett said that the people who developed Brio had also developed Improv for Lotus Corp. Mike believes that the data subcommittee will hold its first meeting in early January. Check with David Jamieson-Drake for the exact time/date.
John Board reported that the Strategic Planning subcommittee had met with Talman Trask and were favourably impressed with his support for ongoing efforts to improve computing at Duke.
Mike Pickett updated ITAC on the status of plans to deal with the problem that the mainframe is operating at near-capacity (95%-97%), handing out a spreadsheet-cum-graph comparing *estimated* costs, over a five-year period, of the mainframe-extension options listed below. The five-year period was selected as a hypothetical point at which current systems running on the mainframe might all be moved to other platforms; please note that there is no firm date set for the accomplishment of that task at this time. Also please note that the dollar estimates in the spreadsheet are just estimates, and are not based on any firm bids or quotations received from vendors, etc.
The options compared (against a baseline of no change) were:
- Upgrade the existing system by adding an additional processor
- Purchase a new CMOS parallel-processing system to replace the existing system
- Lease a new CMOS system to do the same
Some of the many considerations discussed that need to be factored in include: The punitive price structuring of support and maintenance of the existing system used by the vendor to push users towards reinvesting in newer systems, coupled with attractive packaging of of the new system with very low software costs; the amount of labor involved in bringing up and becoming familiar w/a new system vs. simply dropping an additional processor into the existing one; the amount of space that could be reclaimed by replacing the bulky existing system with the smaller new one; the relative ease of financing the processor upgrade vs. the lease or purchase options (as stated by Betty); and the possible extended usefulness of a new system, should we purchase one, for other server-type tasks after the existing mainframe applications find new homes.
Charlie Register reported on the results of his investigation into the costs and difficulties associated with moving a specific application, the Advance system, from the mainframe to a new platform. He looked into two platform options, neither of which is available yet except in Beta form; and he concludes the transition will be expensive and difficult, with software costs alone in the 250k range and total costs, including conversion and hardware, perhaps in the neighborhood of a million dollars.
Mike reported that some of the previously discussed "tweaking" of apps. on the mainframe was performed, and appears to have been successful enough to get it through December without a major failure of some type. He said another way to reduce the load on the mainframe in the short term, should the situation become critical, might be to move the Advance application to the Medical Center's mainframe. Mike mentioned that there could be implications on Medicaid / Medicare reimbursements if this option was pursued.
Another short-term possibility for extending the existing system might involve using some sort of high-end Cisco router product. Bob Currier suggested that Federal Express had good results moving a large portion of their TCP/IP load off of their mainframe. However Mike and Charlie Register agreed that we were already doing about as much in that area as possible with the TCP/IP load on the mainframe. Since Bob Currier was not present, the possibility remains that he might have been referring to some recent technological improvements that could make further reduction of the TCP/IP load possible using the Cisco product in question, but this is speculative.
Mike noted that Tallman Trask supported the idea of setting a reasonable target date for moving applications off of the mainframe and working backwards from that date. There was some discussion of the pros and cons of such an approach. In at least one other institution, the word is that the setting of an unrealistic target date, coupled with lack of consultation with the affected parties, resulted in the resignation of several high-level IS staff members. On the other hand Betty pointed out that the farther in advance staff knew about such phase-outs, the more time they would have to position themselves to meet the changing job demands that would result. Other institutions have noted that the cost of moving applications off of the mainframe can be substantial.
Another important factor is how long it takes to move the current applications off of the mainframe. Using the hypothetical five-year figures, if the task were accomplished ahead of schedule, the lease option looks most attractive financially; but if it takes longer than the scheduled time, the purchase option looks best.
Betty brought up the question of whether or not Duke should have an official online policy / guidelines, and if so what role ITAC should have in its development. Kathy Underwood has been gathering information on the topic, including examples of other institutions' policies / guidelines. Betty met with Judith White, Myrna Adams and Kathy to discuss Kathy's findings; according to Kathy's research current law on the matter is "all over the map". A general discussion ensued about whether current harassment policies, etc. were sufficient, whether a specific set of online guidelines should be developed, or whether users should simply be reminded of existing policies in the context of online communication. The idea was raised that "less might be better" with regard to specific policies in this area. Mike Pickett noted that, in the absence of a specific policy, OIT staff were sometimes in the position of formulating policy-like positions ad hoc. As an example he pointed out that the Duke webmaster works for OIT and decides what pages are linked to from the Duke home page, as well as who can store things on the Duke webserver. Mike would like to see some external input / oversight on those sorts of decisions. Consideration will be given to the formation of an ITAC subcommittee to look at these issues more closely. Marion Shepard suggested that the membership of DOIT (Duke On-line Information Team) be opened up.
Betty reminded everyone that the Capital Campaign is entering its "silent phase", wherein Duke will be approaching corporate donors, who often make donations of equipment. If anyone has equipment needs for a project with the sort of program content that might appeal to a donor, please write up a short description of the project which Duke may present to potential donors as a suggested donation option.