DUKE ITAC - January 30, 2003 Minutes
Minutes: January 30, 2003
Attending: Ed Anapol, Mike Baptiste, Pakis Bessias, John Board, Dick Danner, Angel Dronsfield, Brian Eder, David Ferriero, Nevin Fouts, Tracy Futhey, Patrick Halpin, Alfred Trozzo (for Paul Harrod), Robert Zimmerman (for Scott Lindroth), Roger Loyd, Greg McCarthy, Melissa Mills, Bob Newlin, George Oberlander, Lynne O'Brien, Mike Pickett, Rafael Rodriguez, Molly Tamarkin, Fred Westbrook, Robert Wolpert, Steve Woody.
Guests: Bob Currier, Debbie DeYulia, Dan McCarriar, Chris Cramer, Dave Kass, Dana Risley, Jen Vizas.
Call to order: Meeting called to order 4:04 pm.
I. Review of minutes and announcements
Robert Wolpert welcomes everyone to the first meeting of 2004. He reminds members that minutes from the last meeting were sent out last week for review. Please return any corrections or comments to Mike Pickett.
Michael Gettes hired
Mike Pickett announces Michael Gettes has been hired. Mr. Gettes will act as Mike Pickett’s counterpart, affecting a service pull and technology push. Mike Gettes will focus on what OIT does with technology as we move into the future. He will start in early April.
Mike Pickett announces the Duke University home page is ready for testing. There is a link to the new page at the top right corner of the existing page www.duke.edu. If you find something that doesn’t work or is incorrect, send comments to the link on the page.
High performance computing monthly lunch
Mike Pickett announces plan to get people together who have interest in high-performance computing (HPC). The meeting objective is to discuss problems and solutions regarding HPC. The first meeting will be held in LSRC D106 on February 10. If you know anybody intereted in HPC, please refer them to Mike Pickett and he will get them on the invitation list.
Tracy Futhey: Tallman Trask sent out a memo at end of 2002 announcing a reorganization in his area resulting as a result of the retirements of Jerry Black and Joe Pietrantoni. One of the changes relative to technology is the Duke Card office going to Mike Mandl's Financial Services area because of the financial transaction involved; however, technical oversight of the Duke Card will be under OIT.
The card office processes a whole lot of money so it is better that the Duke Card be associated with Mike’s office. We have talked about how we can leverage the Duke Card more and there is room for that in the future. Another change affecting OIT is the publications function coming under OIT.
Tracy Futhey: At the last meeting we discussed establishing two new subcommittees, and reviewed the charter of the network subcommittee. (View the network subcommittee charter at http://www.duke.edu/~picke001/ITAC/Portal_Charter_v2.1.html.) The second committee is exploring what we do with portals. (See http://www.duke.edu/~futhey/itac/portal_sub.htm.) That group is meeting in a few days. These two subcommittees take the place of what had been the video subcommittee and the content management subcommittee.
Common Solutions Group (CSG) Update
Mike Pickett: The meeting was held at the University of Texas in Austin. I found it interesting how many people thought there was a difference in culture between IT people and library people. It was interesting to hear what others are doing, particularly MIT with their D-space initiative. We talked a bit about digital rights management with materials, for example, what happens when materials are online? We also talked about how it affects libraries, and we discussed issues of copyright. We went through a discussion of ways to maintain digital material. In general the sense is there is a lot of work to be done in this area still. Paul Conway led a discussion on Duke’s library efforts and interacting with IT.
David Ferriero: I was one of three library directors there. It reminded me how close we work together and how distant other universities libraries and IT groups are.
Mike Pickett: The next day was spent on attacks on content providers. There was a lot of focus around authentication, and Shibboleth. I was proud about what we have done with Shibboleth. It is in a very mature stage. Getting it there was a very complicated task and this speaks well to where we can go with collaboration.
The regular meeting consisted of two workshops. One dealt with bandwidth. We talked about p2p networking and different ways universities have approached the problem. Two places that really shined were Terry Graves of the University of Washington. We have a relatively strong way to approach the problem in terms of throttling down the Resnet volume, but some institutions can’t separate traffic like that. No one recommended shutting them [p2p applications] down completely. We are going to have to be vigilant. Cornell had a rate-based billing where if you suck up bandwidth, you pay for it.
Tracy Futhey: The modeling by folks at Cornell suggests that making students buy bandwidth results in consumption dropping dramatically. They provide a basic bandwidth per month and if you go over that you have to pay for it. Another interesting discussion was spam and what to do about it. The University of Minnesota has an approach that almost eliminated all spam. First, they don’t accept mail from known spam areas; Second, they use Spam Assassin and other programs to filter spam out. The way they deal with exceptions is to send an automated reply, a bounceback, to the message originator. Most spammers just drop the address off their list. If the originator of the message is not a spammer, but is a legitimate person, the real person reads the reply and sees that their message did not make if through the spam blocking measures. The message basically says, “If you think this is wrong let us know,” and then the original message is re-sent with a wrapper around it that identifies it as possible spam. I shared this with Chris and Rob, and Micheal Gettes was also at the meeting so I hope that is something we can talk about again soon.
Mike Pickett: They [U of MN] said they were glad to share the technology with us, but their code was still messy at this point. We also had a long discussion about Chandler. Chandler is an e-mail MS Outlook competitor supposedly written so it will be polite and live on your desktop handily but will also follow some standards. Several groups spoke with the developers of Chandler and offered advice on how to make it fly in university environments. There was sort of a rousing, “lets go!” But there was a, “lets go carefully forward.” So for right now we will keep watching Chandler.
II. Shared high performance cluster computing proposal
Presented by Bill Rankin
I work for John Harer at the Center for Computational Science, Engineering, and Medicine (CSEM), and we are proposing a shared cluster on campus—a moderate- or mid-sized cluster—used as a shared resources by several researchers on a buy-in approach. In many cases there are more scientists using clusters more and more every day. A lot of clusters are hard for researchers to put together and administer so they can be used effectively. One idea is to put a cluster in a shared facility space. Groups using the cluster will share sustained costs and in exchange researchers get dedicated use of X percent and shared use of the rest of cluster. By co-locating and administering these clusters as one, we can provide services like that. We have some initial funding and we’re looking into research groups putting something like this in their grant proposals.
Melissa Mills: John Harer gave a proposal of people buying their own machines and having a place to house them. Are you thinking of including a possibility for people who have their own machines?
Bill Rankin: Those ideas are still on the table. One facet is the service center approach were researchers provide funding and we provide purchasing and maintenance. This was a simpler approach so we went with this first.
Tracy Futhey: We knew we had interest from some people, but more were interested in this approach so we went this way first. We had identifiable near-term customers.
Molly Tamarkin: Do you have a provision for people to buy in for a short-term commitment? Say a year?
Bill Rankin: Yes, we are looking at those models.
Molly Tamarkin: who manages this? Who is this under?
Mike Pickett: This is sponsored by the Provost’s office.
David Ferriero: Where is it—what is the status— and what is OIT’s role?
Bill Rankin: OIT wants to enter a period of negotiation.
Tracy Futhey: John and I have been talking about this since my arrival. OIT will be involved, not much in the beginning, but over time there are probably appropriate spaces to have a role, but none are resolved at this point. When it becomes an infrastructure service we’ll get involved.
Robert Wolpert: It seems scheduling will be a bigger headache than we think. Is there some shared structured way to allocate time?
Bill Rankin: UNC uses a job schedule with different levels of queues. If you pay for a larger percentage of time, jobs you submit go to a higher priority queue.
Robert Wolpert: What do you do when machines get old and no one wants them, but you are heavily invested in them?
Bill Rankin: Our financial model accounts for a three year refresh rate so machines should not get old.
John Board: MCNC is also setting up a cluster operating on free model. I think the North Building is place to start ours. There are probably a lot of server shops in RTP that we could negotiate terms on. We could negotiate for space or perhaps people. MCNC has a bit of empty floor space now.
Melissa Mills: Do you need systems level access to configure applications so they do the job? How to you handle that?
Bill Rankin: It depends on the case. Application specific issues need to be dealt with directly by the research group.
John Board: We may be able to effectively refer people to MCNC for training, which they are very good at. We don’t want to replicate things they are doing out there.
III. ITAC Academic Technology subcommittee report: Technology in the classroom
Presented by Mike Pickett
View the ITAC Academic technology committee's report on classrooms at http://www.duke.edu/~picke001/ITAC/class_tech_12-16.html
Additional background information and minutes from the Academic technology committee meetings can be found at http://www.oit.duke.edu/itac/acadsub
Mike Pickett: The committee was put together in the fall as a successor to CITIE. The intent is to bring in people with an overlap to ITAC. A predominant number of people were faculty with teaching awards. We also have a student representative and the vice president of academic affairs. We’re looking for specific recommendations about what we do in our classrooms. Jen Vizas and Sarah Roberts helped a lot in this effort.
The most important thing this group came out with is the difficulty in getting policy and practice together. For example, we have a language class we need audio for, but the classroom doesn’t have it. There are scheduling issues we need to think about improving. There is a committee looking into scheduling classrooms and spreading technology around campus.
We talked a lot about support and what faculty need while teaching. They want an immediate 3-5 minute response time in a classroom situation. How can we prevent a need for these immediate responses? Can we do periodic pre-checks? Scheduled maintenance just like on a car? Melissa did a card where people can give feedback. Another issue is how do you cycle technology? It would be nice for faculty if they know how things work.
An interesting idea is mobility. Everything doesn’t need to be in one classroom. Projectors are now getting so cheap that we can buy several and shuffle them around.
We’re also considering an online Web course request form. Faculty can ahead of time go through a checklist of what every course needs: projectors, online reserves, u-style vs. tier style classrooms, etc. This info is populated in a database. Then all the different departments can use the database to shop around for the technology and equipment they need.
Melissa Mills: Bob Thompson asked me to bring to this group his strong endorsement of the idea that we get the scheduling in order. If not, everything after that is dead in the water.
IV. DSL services update
Presented by Angel Dronsfield
Two documents were endorsed by groups of staff faculty and support folks after we presented them with four findings. We met with a series of vendors. We recommend two very good solutions for remote access that we think are better than the current Duke DSL option.
Dan McCarriar: We will be able to offer much higher speeds through these options. There are two methods of billing. Individuals can be billed directly or Duke will get one bill for everyone.
Robert Wolpert: I heard that static IP is not necessary if we get VPN together.
Dan McCarriar: Yes.
Tracy Futhey: Thanks for everyone’s help and Dan especially for coming in as the new guy and living through this project.
V. Proposed letter to students re: peer-to-peer/copyright
Presented by Debbie DeYulia, Bob Currier
Debbie DeYulia: Students had to register their computers with the network at the start of semester. From the help desk perspective it was going fine in December. When students came back we got a lot of calls. Most just required changes to DNS configurations which they hadn’t done before. Next year will be easier because only freshman will be new to the process. We used to require DNS to be configured. Now it cannot be configured for this to work.
Chris Cramer: If anything we have removed permanent configurations.
Bob Currier: From the technical end it went very well. We had a few concerns, but we tracked them all down quickly.
Molly Tamarkin: Did you update the page or just tell everyone to restart their computer?
Chris Cramer: It gave them the options to restart or wait two minutes and renew their lease.
Tracy Futhey: Soon after we had an article in the Chronicle about this. This was relatively easily accepted from the policy side?
Chris Cramer: We did an interview with the Chronicle before this was turned on. Students understood that we have a large number of hacking and virus issues, and we need to get to students fast in certain circumstances.
Melissa Mills: Is this something we can borrow the expertise and install in our area too? We do this for classrooms and so on. It would be nice to use this technology.
Chris Cramer: Certainly. We will write up a page about what we did. It is a fairly simple process.
Mike Baptiste: We are in the process of setting something similar up.
Molly Tamarkin: This is great because it points out how using the network is a privilege. We are planning to use this to track rogue users better.
VI. VPN status udpate
Presented by Bob Currier
We’re live and tested and ready to go. We successfully connected two line connections and a smattering of test users, and we’re looking for a few more. Send Kieth Gentry an e-mail (email@example.com) and we’ll get a client to you. The last step is just getting together with Debbie’s folks and getting the client documented and out there.
Mike Pickett: I can see a lot of people saying, “Why would I want vpn?”
Chris Cramer: There are a couple of reasons. Having a Duke IP address is one. Vpn gives you the appearance of having a Duke IP address and lets you into Duke-related areas. Also, vpn provides encryption so you have a reasonable expectation that connections are secure between home and Duke.
Bob Currier: And the LAN-to-LAN mode allows us to connect with departments using certain applications and use the Internet as the backbone.
Robert Wolpert: What about wireless?
Chris Cramer: Everything over the wireless network is unencrypted anyway so if I have a vpn client on a laptop, I might choose to use vpn within Duke just for the encryption.
Raphael Rodriguez: I understand that vpn doesn’t work with satellite dishes to communicate. We have some people in the Med Center for whom this is an issue. If you have people using satellite they need to be aware of that.
Mike Baptiste: You aren’t restricting split tunneling are you?
Bob Currier: Yes. No split tunneling allowed.
Chris Cramer: To get access to some of the library databases you need the appearance of coming through Duke and split tunneling would not allow that.
Tracy Futhey: Where we’ve gone is proof in concept—we can do this—now we need to go to the documentation and then where we want to apply it.
Melissa Mills: We need some sort of rollout plan to help everyone get up to speed.
VII. Other business
Tracy Futhey: We will soon be ready to talk about new cell phone contracts. If you need one now you can go to the Web site (http://www.oit.duke.edu/televid/basicservice/) and see phones but not rate plans. Ginny will have more information at the meeting in two weeks.
Mike Baptiste: I have a question about the new cellular tower that went up. Can other carriers use tower?
Angel Dronsfield: We are working with Alltel to provide campus service. There is talk to put towers up on the clocktower quad. The Verizon tower you’re speaking of was not something that Duke sought. It was disguised to look like a light on the soccer field and not a cell tower. The result is a good-looking cell tower but an ugly soccer field light. To balance it out we may install another tower at the opposite end of the field.