Duke ITAC - March 31, 2005 Minutes

DUKE ITAC - March 31, 2005 Minutes


March 31, 2005

Members present : Mike Baptiste, Pakis Bessias, John Board, Shailesh Chandrasekharan, Paul Conway, Dick Danner represented by Ken Hirsch, Angel Dronsfield, Nevin Fouts, Tracy Futhey, Christopher Gelpi, Michael Gettes, Daron Gunn, Guven Guzeldere, David Jamieson-Drake represented by Bob Newlin, Deborah Jakubs, David Jarmul represented by Phil Lemmons, Kyle Johnson represented by Tim Bounds, Scott Lindroth, Roger Loyd, George Oberlander, Lynne O'Brien represented by Jim Coble, Mike Pickett, Scott Smith, Molly Tamarkin, Robert Wolpert, Steve Woody, Paul Harrod represented by Alfred Trozzo

Guests: Arie Lewin, Fuqua School ; Randy Haskin, Fuqua School ; Dan McCarriar, OIT; Seth Vidal, Physics; Jess Mitchell, OIT

Start time : 4:04 p.m.


I. Review of Minutes and Announcements:

  • Revised registration windows - Chris Meyer

    Chris Meyer says senior registration for fall 2005 is slated for this Thursday, April 7. He reports having a very successful registration this past fall for spring 2005, and is anticipating a repeat performance of that. There are a few differences this time from last fall. One is that the summer term registration has been decoupled from registration for fall courses; summer registration has already been done. Another difference is that there have been two rounds of functional and load testing, where there was just one last year. Otherwise, there are really no changes.

II. Fuqua Demos: Faculty Research Papers/Seminar Calendars - Prof. Arie Lewin, Nevin Fouts, Randy Haskin

Nevin Fouts says Fuqua is only 35 years old and is already in the top ten MBA programs and top five programs worldwide in Executive Education. Duke is clearly seen as a global leader in management education.

One recent technical development is our teleconferencing system, all of which is online. This is a glimpse into how we might do classroom-to-classroom discussions in the future. In terms of faculty and student advisory committees, there is a faculty tech advisory committee, and we work with a student group called the Technology Advisory Council, which usually consists of five very focused students who are great communicators and are usually not “techies.” We use select students to help us beta test technology. We support faculty with multimedia, research, desktop support, simulations/models used in class, Web space in general, and also help with questions faculty may have along the way after they have started teaching.

In terms of centers at Fuqua, there are a number of centers that are either solely at Fuqua, or in the case of the CIBER, where other universities are involved in the center.

Arie Lewin says CIBERs are a creation of the federal government, awarded competitively for four years. Last time around 30 CIBERs were awarded, and we were awarded for the fourth time. A CIBER has to contribute to the internationalization of their school. At Duke we have outreach responsibilities for the business and other communities around us. We have responsibilities for teaching languages to our students and advancing American competitiveness.

Our CIBER is relatively small, so we have to do more with less. The idea was we would do things that would place our CIBER in the center of various networks. None of this could be done if it wasn't for the technology and the support we've gotten from the Fuqua technology project.

The first project is the InterCultural Edge project. This program has devised a new instrument based on individuals to estimate problem-solving skills, etc. Anyone who wants to do a cross-cultural introduction to teaching can use this program. It is being packaged for use by other schools. We use this to make them aware of cross-cultural differences in situations like communication, negotiation, etc.

A second project is the Offshoring Research Initiative, which tracks companies' experience in off-shoring jobs. The intent is to obtain data from companies from the people who manage off-shored functions. The surveys are run from Fuqua, and it has had an enormous impact in terms of media coverage. We're doing it because all of the discourse about off-shoring is uninformed because we don't know what companies are experiencing. We created a community of companies on a website which we monitor and participate in. For the second phase of the project we are taking it to other western countries facing the same set of issues.

Lastly, CIBER themselves have decided to make use of technology in an interesting way. We looked into developing a CIBER Web at Michigan State University . When we have events to report, we put them onto the CIBER Web site and it populates it onto our site so we don't have to enter it twice. The Fuqua CIBER decided to be the first pilot, and we're well on our way of testing.

Randy Haskin shows some applications to demo this work, Fuqua intranet, home page.

Mike Baptiste asks how many people work with you from day to day?

Randy reports he has seven direct reports and a couple of people who do design with a dotted line to him to make sure everything looks the same.

Michael Gettes asks, what are the types of technical skills of the person who works with you?

Arie says he usually uses a graduate student, and also has a fulltime applications developer, who did the interfaces for both projects.

John Board says he gets the sense that IT is so integrated into Fuqua that it is more regimented than at other schools. What is the balance between telling people what to do and working with what people have?

Arie says they do strive for that balance, and it's amazing how much flexibility technology allows.

George Oberlander asks if they use any particular programming methodology.

Randy says they tried some of that, but it tends to get rigid. Most of what we do is centered around getting requirements right and building prototypes.

Molly Tamarkin asks how do you think you communicate with faculty about what technology they will be wanting in the future?

Nevin says they try to anticipate things that will be enabling for faculty, so there is a proactive aspect to it.

Randy says from his perspective, as a rule of thumb, if it's one specific faculty they generally can't do it. So the wider the availability, they tend to prioritize those.

III. Preliminary Findings: Voice Over IP Working Group - Dan McCarriar

Dan says he would like to acknowledge Nevin Fouts, John Board, Angel Dronsfield, Michael Gettes, and Rafael Rodriguez, among others, who have participated in the VOIP working group. Today he is going to go over the preliminary findings. They aren't written in stone yet, but they do largely reflect what we've been talking about in the group.

There are seven sections in the final report:

1. Overview of telephony at Duke: Discussion of advantages/disadvantages of current environment. Advantages: stable, staff who knows the system. Disadvantages: small legacy systems, many are vendor discontinued at this point; other elements like that the voice mail platform is fairly old; monolithic central office architecture designed to live in one building. We note that while many shortcomings can be addressed with systems that don't have to do with VOIP, the state of the environment now does present interesting opportunities for VOIP.

2. Voice over IP opportunities and perceptions: When VOIP became hot five years ago, when everybody said it would save money when telephony and the Web run over the same network. A lot of people have backed off that stance now. Currently there are two schools of thought. One is that VOIP is a simple telephony replacement, and the other is about taking a bunch of communication devices and linking them together, enabling new applications. We believe any transition Duke takes to VOIP will be an incremental one. People will need to get used to the idea of an integrated communications platform rather than just a phone on their desk.

3. VOIP marketplace and trends: The industry at large is out of the early adopters phase to the early majority phase. We've seen a consistent increase in the shipment of telephony stations in general and the percentage of IP-based stations in the mix has increased. In terms of other universities, they are all over the map. Stanford has committed heavily to VOIP, and Dartmouth has done a fairly large deployment starting with 700 student stations. We also see groups like Internet2 looking at how to provide voice services over their networks.

4 . Requirements for VOIP at Duke: These range from what kind of telephony features Duke would need from security to integration with enterprise infrastructure. The idea was to have something to hand to vendors when we are ready to take next steps. A second document we came up with was more Duke-focused: what do we need to take care of?

5. Vendor landscape: What vendors we might pursue in terms of next steps. When you go down a path like this you look at your existing vendors, and then ask are there other players we need to look at? The answer we came back with is probably not. We have a large number of lines and a large telephony environment. Many vendors create products that serve fewer lines. Avaya, Nortel, Lucent, and Cisco are the primary vendors we would look at.

6. Next steps: There are a number of strategic questions outside the scope of this group. One is the financial model: do we actually save money doing this? To what degree do we want to pursue a "convergence" model – do we want one network that carries everything? To what degree can or should VOIP change the environment from centrally to non-centrally run?

There seems to be general consensus in the group that we need to move forward with vendor discussions and product deployments. We need trials designed to answer some of these strategic questions.

Nevin adds that one key point we're wrestling with is because we agree on an incremental approach you have to think about inter-operable call centers. To stress what Dan said about trial deployments, they are to see how parts work that can connect later.

John Board says the vendors are showing compatibility at handset level, but are almost hostile toward compatibility at the backend, exactly opposite of what would serve us best.

Angel Dronsfield says one consideration we'll have to deal with is how we will talk to the outside world. These systems allow connectivity on campus, but we have to be able to connect outside of Duke as well.

Molly Tamarkin asks if research at Stanford and Dartmouth looked at staffing.

Dan says we didn't get that in-depth with benchmarking, but we certainly have contacts there we could question about that.

Robert Wolpert asks, are there 9-1-1 issues?

Dan says it isn't completely resolved, but it is being addressed. There are infrastructure issues that cause some problems for databases that would support 9-1-1 (e.g. wide access to wiring closets).

IV. Linux@Duke: The Process for Considering Consolidation - Mike Pickett, Seth Vidal, Jessica Mitchell

Mike Pickett says from some of the things we were looking at for an OIT/Arts & Sciences collaboration, this is one thing that could be a good collaboration. Linux@Duke grew up as a backroom, furnished with donated equipment from Physics, Math, etc. Now it is comprised of 1,200 machines around Duke and the world. Seth has developed something called Yum (Yellowdog Updated Modifier) that has been picked up by Red Hat. We've gotten a lot of benefit from a relatively low contribution, but because it has become so successful we need to formalize it more. Luckily none of these machines has been hacked in four years, but who knows if they will continue running in six months. A lot of people in this room have a stake in this, and we have contributions from people all over campus. We want to talk about the process, what we're thinking about doing, and get your input and feedback. We came up with a rough idea of how it might work, and then Jessica talked with other people and wrote up proposals on how it might work. I haven't found anyone who says, “This is a bad idea.” We have some tentative approval to go ahead.

Seth Vidal says currently they are providing services a lot of people are probably using like update mechanisms for systems, local supply for most Linux computing needs so we don't have to use offsite bandwidth unnecessarily and get support we have to pay for from other organizations. Another service they provide is venues for folks on campus to interact. In addition, we've been providing some testing area to roll out things clearly not production ready. A transition or consolidation would provide us with more time to investigate these things so we could possibly transfer them to enterprise level. There is a community component to linux@duke . At Duke, for this area, because of our available resources, there is a responsibility to help out other universities and folks who aren't as enabled. We make it possible for them to do things they wouldn't otherwise be able to do. We make programs available for anyone else to use, and anyone else to improve. One advantage is that if we contribute, we get a seat at the table. When things come up for changes, we can impact the decision for technology development at the higher level.

Jess Mitchell says after the meetings Mike mentioned, she sat down with Seth and listed services and each machine with specs and their current status. This group has worked with no replacement schedule or budget, and is mostly worked on by volunteers. She's split the machines into critical and non-critical, and cost evaluation was based on if a machine was mission critical. Start-up would be $16,000; based on a three-year depreciation schedule; after that we are looking at about $10,000 each year not including employee salaries. We are thinking staff would be two people, one full-time and one half- or full-time. The needs of the group are structure, a formal budget, formal employees, and machines. Then we split this into three options, keeping in mind linux@duke is two-fold: the services and the innovation wing. The services offered are concrete, whereas the innovative wing is more the community. The three options we considered are 1) linux@duke stays in Arts & Sciences, 2) it is shared between A&S and OIT, and 3) it is moved exclusively to OIT.

Option 1) is not very feasible because the machines are too old. We could do something like the following: machines and personnel move to OIT, the budget could be shared, namely the employee salaries. Whether it moves to OIT exclusively or not, the needs are the same. What remains seems to be largely logistical issues and an assurance that the innovative part of linux@duke would remain.

John Board asks is it clear that the best option is one full-time employee, or would two part time be better? He expresses concern about someone not staying in touch with demands of production and everyday systems.

Jess says the person would be surrounded by people working on the systems, so that shouldn't be too much of a concern.

Seth says the reason we want two is to have a backup, we need to have something there.

Michael Gettes says in the current model linux@duke is supporting 1,200 machines. With the new model is there any sign we'll bring in more?

Seth says there were 800 machines in 2004, so the number has grown significantly in the last year. A lot of this is attributed to clusters, but he doesn't think we'll see a decrease in that number any time soon. Having these services better provisioned isn't going to hurt the spread of Linux on campus.

Mike Baptiste asks why is there a question of where to go in terms of whether OIT takes it on or A&S handles it centrally?

Shailesh Chandrasekharan says he is a faculty in physics, and a lot of money in development has been saved because of linux@duke . I think if we had to manage this by ourselves and pay someone to look after this, there would be less money towards research.

George Oberlander says it looks like there are two issues of production-oriented support as well as the foggier innovation, and maybe some uncertainty as to whether one organization is best suited to handle both needs. Maybe the model for innovation is one the open-source community at large uses, which is volunteer-based.

Jess says the idea was for OIT to provide the infrastructure for the university. The innovative part is the community; it can't become OIT exclusively.

Mike Pickett says part of what we need is a formal structure to make sure that system doesn't break.

Molly Tamarkin says although this was developed in A&S, other schools rely heavily on this service, and we can use the opportunity to get more feedback on what an expanded service would look at.

Paul Conway says he thinks this is a commendable project; there are a lot of questions out there about cross-departmental collaboration, and I think we should continue to do this kind of cross-talks.

V. Other Business


End time : 5:34 p.m.