DUKE ITAC - November 5, 1998 Minutes
November 5, 1998
Present: Ed Anapol, Landen Bain, Pakis Bessias, Chris Cramer (for John Board), Kevin Cheung, Ken Hirsh (for Dick Danner), Jim Dronsfield, Jim Coble (for David Ferriero), Patrick Halpin, David Jamieson-Drake, Melissa Mills, Kyle Johnson (for Caroline Nisbet), John Oates, Mike Maupin, Rafael Rodriguez, Leslie Saper, Bill Scarborough, Marion Shepard, John Sigmon, Clark Smith, Robert Wolpert
Visitors: Ginny Cake, Bob Currier, Charles Register, Phil Verghis
Review of Minutes & AnnouncementsThe minutes for October 22, 1998 were approved without any modifications.
Robert Wolpert introduced Patrick Halpin as a new member of ITAC. Patrick Halpin is a faculty member at the School of Environment.
Robert announced that Betty would be back in a couple of weeks after her honeymoon in Europe.
Follow-up discussion on United Way mailing listRobert Wolpert opened up the discussion on this topic by announcing that it was a follow-up discussion to a concern that had been expressed at the October 22 meeting during the discussion of the appropriate use of mailing lists.
Rafael Rodriguez reported that he had received an e-mail from Bill Scarborough stating that he had been handling the United Way data for Duke University for the last three years and to the best of his (Bill Scarborough's) knowledge United Way doesn't receive Social Security information. Rafael mentioned that he had contacted Marion Shepard with this information and that Marion was going to follow up with his contact.
Marion Shepard distributed a copy of the list he had been provided, with sensitive information blacked out. Marion had been informed that the list had come from United Way.
Bill Scarborough mentioned that he recognized the report since it's one that he produces. Bill stated that the information is not provided in that form to the United Way. Bill stated that he receives an extract from the payroll file consisting of name, Duke address, social security number, paypoint, component code, and any current payroll deduction for United Way. Similar information is sent to the Publications Group to print the pledge cards. To Bill's knowledge, United Way itself does not receive this information. Bill uses these data to track donations by employee, paypoint, and component. Bill reports summaries to the campaign chairs and United Way itself. United Way only receives summary information during the campaign. When the campaign ends Bill sends new payroll deduction information back to Payroll for use in the new calendar year. Bill also sends name and address (but not social security number, paypoint, etc) to United Way for donor acknowledgements. If the donor has requested anonymity, Bill does not include their name.
Robert Wolpert mentioned that it appears that sensitive information is not being provided outside of Duke but that it seems to be circulating within Duke.
David Jamieson-Drake suggested that instead of using social security numbers that Duke unique ID be used.
Ed Anapol stated that he had spoken to George Kolasa. Kolasa said that the plan was to move to Duke unique ID.
Project Enterprise Update
Mike Pickett gave an update at the board on the key dates for Project Enterprise. Mike stated that the date for the implementation of Financial Information/General Ledger (FI/GL) was moved from January, 1999 to March 1999 due to the number of people that have to be trained, the large number of interfaces that have to be tested, and the number of operational changes. Mike stated that it's key to make the March date since we have obligations to Durham Regional Hospital and Raleigh Community Hospital that could be affected if the dates move further.
Mike mentioned that FI/GL is being implemented with SAP Version 4.0B. Currently the Procurement system on the campus side is implemented on SAP Version 3.1H. Procurement will be moved to SAP Version 4.0B on the campus side in March, 1999. Additionally, there is an alpha version of SAP (Version 4.5A) that is being used to start the configuration for HR and payroll. However, this version cannot be used for production since it doesn't have the tax tables. SAP Version 4.5B will be the official SAP version that has the appropriate tax information and additional changes that Duke has requested. The first application in HR that will be turned on is the recruitment system. Web access will also be activated in the first half of 1999. Mike stated that this is the foundation and that significant work lies ahead. The summary of the dates as presented by Mike Pickett is as follows:
Discussion on guiding principles for cluster equipment upgrade
David Jamieson-Drake introduced this topic. David mentioned that we want to do the upgrade of the clusters as wisely as possible and that we should have a discussion on the principles to employ to turn over the cluster equipment.
Robert Wolpert stated that the purpose of the discussion was to identify who has a stake in the clusters and what should be the guiding principles. Robert mentioned that ultimately the decision is for Betty to make but that she seeks ITAC's advice. Robert opened the discussion by asking what are the "decision drivers."
John Sigmon stated that last time we talked exactly about this matter and that a decision had been made for a multimedia lab that the students didn't like. John stated that we should look at the minutes of that particular ITAC meeting and identify what didn't work.
Mike Pickett mentioned that the process last time had been different. A solution had been proposed and there had been discussion on how the ITAC community felt about the proposed solution.
Robert Wolpert said that we didn't follow this approach on the previous discussions and that we hadn't taken into account students as stakeholders.
David Jamieson-Drake mentioned that as a researcher he would like to approach it from what statistics we can gather. We should find out the student preferences from what they bring to campus, and from student usage of the clusters. David volunteered to go to the board and write down the key criteria (stakeholders and guiding principles).
A discussion ensued on the student (undergraduate and graduate) as being a key stakeholder. We should look at survey information on what students bring to campus, cluster utilization (number of seats filled, programs used, bandwidth, storage, etc.)
The role of the clusters came up. An argument could be made that the clusters should complement what the students bring to campus.
Chris Cramer suggested that faculty preference needs to be taken into account. In order to grade a computer science course, he needs all students to use the same equipment.
John Oates stated that he needs to know what is available to the students.
Kevin Cheung said that location is a key parameter.
Clark Smith asked that the grad students be included in the location discussion.
Robert Wolpert stated that we not only need snapshot information but trends-industry and environmental trends. Robert also stated that it had been said at one time that the clusters were for undergraduate students in Arts & Sciences and in engineering. A key question is what is the mission of the clusters-whom do they serve? Robert also stated that the percentage of students who bring computer to campus is rising and that may argue for a complementary role for the clusters. Additionally, many companies make donations so we may want to take into account philanthropic donations as a consideration.
Mike Pickett mentioned that we should consider the role of kiosks for web access and e-mail thus freeing the cluster equipment for a more compute-intensive role.
Kevin Cheung requested that in addition to cluster equipment we should take printers into account.
Clark Smith said that clusters also play a role in research.
John Oates brought up the fact that there are other clusters besides the one that OIT pays for and asked whether we should take those into account.
Kyle Johnson mentioned that the cluster that student affairs set up was to provide a function that the other clusters didn't provide.
Robert Wolpert observed that there are hidden subsidies if departments provide their own clusters.
Melissa Mills said that she thought that this matter was being approached too monolithically. Melissa mentioned that there are multiple needs and that we shouldn't be applying the same principles to all the needs.
Robert Wolpert said that a monolithic solution was not being proposed.
Phil Verghis mentioned that some of the clusters are used for classes and that there are support implications that must be taken into account. For instance, Kiosks break down more frequently than desktop computers.
Jim Coble stated that electronic resources provided by the library had printing implications and that instructional technology initiatives will drive needs.
Patrick Halpin said that some clusters need a high degree of homogeneity while others permit a heterogeneous environment.
John Oates mentioned that he felt that last year there was an undertone that clusters should be phased out.
David Jamieson-Drake said that the Provost considers that clusters are a key since we don't require students to bring their own computers.
Robert Wolpert asked what data we should try to get.
David Jamieson-Drake answered that we should get the utilization of seats and programs without having to send people around to the clusters to collect the information.
Robert Wolpert asked what was possible.
Rob Carter answered that recently metering software had been installed that provides information for PCs, Macs, and UNIX systems. Rob mentioned that is more difficult to get the seat utilization on PCs and Macs since the users don't have to log in on those systems. However, that information can be inferred by using screen saver statistics.
Robert Wolpert asked if Rob Carter could accumulate information on the UNIX, PCs, and Macs and come back in two weeks and tell us what can be provided.
Ginny Cake asked that the information be accumulated by location.
Patrick Halpin warned that utilization can be dangerous if used by itself since the clusters may not be configured appropriately for some users.
David Jamieson-Drake asked if there was a need for a survey.
Robert Wolpert asked if we would be getting information that is different from what can be obtained from other sources.
David Jamieson-Drake said that we will get to users whose needs are not being met. Additionally, other stakeholders such as faculty and student service providers could be surveyed. He asked if it would be helpful to send a sample survey out.
Clark Smith said that from his standpoint it was more important to determine the role of the clusters.
Robert Wolpert observed that we had exceeded the time for this topic. He asked Rob Carter to come to the next ITAC with the information that he was able to gather.
The notes written on the whiteboard by David Jamieson-Drake follow:
Stakeholders "Decision Drivers" -> Cluster Purchases
<li>Student Preferences -- undergraduate and graduate/professional
-- surveys of what's being brought and bought
-- cluster utilization
</li><li>"Complimentary" role of clusters
-- don't reproduce what's being
brought by students
</li><li>Faculty + instructor preferences and needs -- highest + LCD
</li><li>Dealing with obsolescence
</li><li>Location location, location -- student lifestyle (ugrad/grad)
also *what* is being done -- software usage information
</li><li>Not just snapshots, also trends: industry, environment
</li><li>Whom do the clusters serve?
</li><li>Gift opportunities/ external "partnerships"
</li><li>complementarily with school/departmental clusters
-- hidden subsidies
</li><li>No monolithic solution -- need multiple approaches
</li><li>Classroom facilities -- clusters used for classes
</li><li>Electronic resources provided by library and University resources
+ printing implications
</li><li>Instructional technology initiative will drive needs
</li><li>Departments [ed note: with content to provide]
(e.g. student affairs)
</li><li>Some clusters need homogeneity, others permit heterogeneity
Bob Currier was asked to attend the meeting to give an update on the network (DukeNet).
Bob presented information on the current backbone and stated that it is currently a mix of FDDI, Fast Ethernet, and ATM. The hardware used are two 7513 Cisco routers, and a mix of other Cisco equipment that together perform about 80 percent of the work. Probes on the network show a utilization of 8 to 12 percent (average) across the FDDI ring. Most segments show 5 to 15 percent average utilization. The routers are running at about 25 to 50 percent CPU load. The internet load is averaging 12 Mbps, and the maximum observed is 35 Mbps (the sum of incoming and outgoing traffic).
Bob observed that currently the FDDI ring is not overstressed and that the Gigabit Ethernet standard was recently ratified. Bob said that prices are falling on Gigabit Ethernet, layer-3 switching, and other hardware. Additionally, there is a trend to move the network layer to the hardware (silicon). Bob recommended that we should keep a close eye on the network usage and be prepared to move when necessary. Bob mentioned that his group is bringing up an MRTG page for all the routers viewable by the community. This page will keep the key router statistical information. Bob stated that he was looking for community input and that we should consider having a similar discussion to the one on the clusters identifying the key guiding principles.
Robert Wolpert expressed a concern that historical usage might mislead us. The applications being deployed might drive usage up significantly.
Bob Currier tated that the routers we have can be upgraded.
Robert Wolpert asked how long if would take if a decision were made to move to Gigabit Ethernet.
Bob Currier answered that it would take 30 to 45 days and most of time was driven by vendor delivery schedules.
David Jamieson-Drake brought up the fact that a couple of new applications being deployed include document images. Both admissions and financial aid are moving to applications that transmit student submissions (applications) over the network. David mentioned that this volume is highly seasonal and some of the information is very graphical and include TIFF files.
Mike Pickett asked if there was any information on the peak utilization of the network.
Bob Currier answered that the maximum peak observed was 75 percent.
Melissa Mills asked if there were any simulations that could be done that will give us some indication of what could be expected.
Bob Currier answered that his group can work with different groups to get a handle on the network load.
No other business.