ITAC Meeting Minutes - February 08, 2024
Allen Building Boardroom
4:00 - 4:05pm: Announcements (5 minutes)
Victoria Szabo: Approval of 10/26/23 minutes.
Tracy Futhey: Resharing the news that Nick Tripp has been named Interim CISO. Most of us know Nick as a frequent host of “Scary” ITAC. Nick was Deputy CISO and helped to build the current team alongside Richard Biever. We had many qualified candidates from outside, but we believe Nick as Interim is the best choice at this point.
Tracy Futhey: The Research IT Needs Process is now being called “Research Needs” to encompass all different aspects. We are moving forward with positions within OIT and Duke Libraries. The Office of Research & Innovation (ORI) is deferring their positions until others have been put in place. The Deans will be updated at the Deans Cabinet meeting, in two weeks and University Priorities Committee (UPC) in March. Jenny Lodge met with some of the research Deans earlier today. Members of the implementation teams were there to participate in the discussion. We have engaged an outside communication group to develop communications strategy for the effort. Several faculty are involved, including Sharon Gerecht (Pratt) and Jerry Reiter (Chair of Statistical Science). We have great stakeholders involved and will ask ITAC members for help on messaging.
4:00 – 4:30pm – Special Guest: Vice President Frank Tramble,
Communications, Marketing, and Public Affairs (30 minutes)
What It is: Mr. Tramble will present at ITAC to share his vision about Digital Communication at Duke and his thoughts on how social media can be used in the context of education and research.
Why It is Relevant: Mr. Tramble leads the university’s central communications, marketing, media relations and brand management activities and will provide additional leadership for communications professionals working in schools and units across the university. Social media plays an outsized role in the day-to-day life of many students. ITAC members will learn how social media can be effectively leveraged for education.
Victoria Szabo: Introduction of Frank Tramble, Vice President for Communications, Marketing and Public Affairs who will give us his vision for Communication at Duke.
Frank Tramble: My focus today is on Digital Communications in Educational Research. Frank has been at Duke for seven months. Right now, the biggest part to digital communications is thinking audience based, how are people consuming information. The rise in digital communications has become more complicated with the number of apps and channels available. The space where we talk about how we show up in every single space vs. how do we make sure that we are pushing audiences to places we want to is a big part of our strategy. Previously the team was focused on how to deal with news media. The journalism space has changed and the focus is no longer on a journalist picking up a press release.
For the educational side, I am meeting with Professor Aaron Dinin, who teaches about entrepreneurship and social marketing, to talk about social media. I see social media as an avenue that continues to grow. Many universities will continue in this space.
Additionally, we are looking at taking some of the content that may have traditionally been in books and digitizing them. We are making this an interactive conversation. This is how the students want to learn. Education is being transformed by how the students consume information. How the students function now is different from the past. They want AI and digital tools. They grew up learning from technology and social media. Someone born in 2003 is now a 21-year-old.
Tracy Futhey: We do have student members of ITAC and are interested in what they think.
Andy Li: As a 2003 baby, I definitely agree.
Frank Tramble: We are looking at our audiences and how they consume information. That’s how we are driving our thought process for digital communications. For example, should Duke be on TikTok? It depends on how our audience consumes information. If they get information through TikTok, then we need figure out how we use it to get our information out. Or we have to do a good job of marketing to people to get them to our platforms. What we can’t do as a general philosophy is believe that because we want something in a particular space, that everyone will adopt to it. We need to have a focus.
We are also looking at the decentralization of our sites and storytelling. We want to know what stories are told on the multiple sites and how they align. We need to look at how people get to our various websites. We will be redoing Duke.edu. What do we want people to do when they come to our website? What will be our overall strategy? We are working on these questions. When you Google Duke, our website is not the first thing that comes up.
What are some of your challenges in aligning digital communications? Do you see duplication in places?
Victoria Szabo: I have a question about when are you speaking for the institution or when are you speaking for yourself when you are in a social media context? What are your guidelines in how you should do that?
Frank Tramble: We have to make decisions on whether you really need a social media account or something else. Is there a better opportunity to align somewhere else? Many people want a social media platform, but they have no strategy or plan in place for it. Without a plan there will be little benefit. Duke Today and Duke.edu are not servicing people well. At Howard University there is a platform called the Dig. It is a central storytelling hub that allows everyone to put stories in and then spread them back out to their own sites. Do we have the right things in place to make it successful? I don’t think we should have a website just because someone wants one. I don’t think we should have social media pages that exist for only one hundred people. What is needed and what is the best way to get messages to the right people? We need a better job of putting guidelines in place regarding social media sites. A previous guideline I created was that you couldn’t have a new site without a six-month plan and a content strategy. Right now, we want to decrease the clutter of all our social media.
Ed Gomes: This is about data and data analytics. How critical is that for you? How do we aggregate that data so that it provides value to our communications team and to you? What is the feedback that you need to help with your projects.
Frank Tramble: Data is very critical and our work is based on data. President Price has charged us to find ways to make sure we are data informed in everything we do. We are looking at brand audits as to where we are and where we want to be. What is the data telling us? Are we in the right space for what we are doing? It’s not just about big numbers, it’s about reaching the right audience. But we don’t know what is working and who our audiences are. We are building out more in data analytics. We are hiring a data analytics management position. We want to look at other departments and see what is going well for them. We want to share knowledge and learn how to we build expertise. We want to look at how we elevate all teams and what tools can be used to make data work better for everyone.
Steffen Bass: Every faculty member is their own media outlet. Some are better at it than others. Can we have training for faculty on how to handle the media landscape? And could we have “dos & don’ts” for administrators? We want to make sure we align with the institution as a whole.
Frank Tramble: That’s a good point. We have several changes happening now. We changed the name of our team to Communications and Marketing. We are hiring three new AVPs to work messaging and media strategy. Part of the strategy will include training the communicators that are already in place. We have instigated a Senior Communications Leaders group that will meet monthly. We have a communicator network that we will leverage more. Trying to align so everyone knows what others know. We just did a media training with Student Affairs. We want to do that for other groups as well.
Tracy Futhey: Duke has a program called Managing at Duke that works with new managers. It would be great to have a Communicating at Duke group training with a focus for the different roles at the university.
Frank Tramble: We want every communicator on all teams to onboard with the central team. We want them to spend a week learning how we work together.
Robert Wolpert: Ten years ago, when we did a refresh of Duke.edu a number of students were unhappy with it. They created their own prototype site. This took several clicks to get to anything academic. Reaching different audiences is a challenge and each group feels some ownership.
Frank Tramble: It would not be possible to create something where everyone is happy. What we care about is that we had the discussion of who are we trying to reach. Who is the audience we are building for? We need everyone to understand that. The front door of the site is student focused.
Robert Wolpert: It does need to be a gateway for academics, students, alumni, and others.
Frank Tramble: We want to know when they get there and how they get there. The challenge is when they click on other spaces, we need to know how built out the other spaces are. As an example, if I am an alumni and their section is not built out, it hurts the experience.
Paul Jaskot: For our program, three hundred hits is a good day. How can we help to let you know that these are good numbers for us and that this does a good service for us? What do we need to provide to you that would help you understand and celebrate that?
Frank Tramble: It’s about intentionality. Did this message successfully lead to some call to action? I spent ten years with Case judging institutional websites. In rating them we were looking at whether they got a result they planned for or did they fall across something that was good? If you are intentional about it, it can be recreated. If it was unintentional, then it is hard to know why it happened. The “why” behind the statistics is what we need help to understand. It’s not just about the fact that you got the clicks, it’s how the audience responds. Did it lead to something that is actionable? It’s about understanding the audience. Where does the story fit in your marketing?
John Board: For the individual faculty websites, with the growth of school level communications staff, we have been told to embrace the brand. Our brand is us. The principal audience for faculty is to attract graduate students to Duke. It is important to have a self-contained view of my work and my lab the way I like it is important to many of us. That is very much a case where institutional priorities of having all our websites look to look the same can very much differ from the actual need, we have to compete for great graduate students.
Frank Tramble: There is still a lot of alignment to be made and this is where the data analytics comes into play. Based off what we know, how can we get people to do the calls to action. If we can agree on what the data is telling us, we can reach our goals. Plenty of professors have websites. It’s less about how the website looks and more about how it functions with the audience. We do want people to align with the Duke brand.
Dave Mc Alpine: Building the brand and using data analytics is wonderful. But the media space moves so fast. How is the communications team preparing for the problems to come?
Frank Tramble: We have navigated this strategically. We need to use everything in our arsenal, including our storytelling capabilities, to affect a message. Part of the crisis strategy is to look at what conversations are happening now and what areas we don’t have defense points. We are looking at what negative stories are happening at other universities. We want to be prepared with articles already written. A good example is the New York Times article about our Pell Grants. We needed to have already formed a statement before the media asked the question. We want to be proactive and understand where we are vulnerable. It all goes back to authenticity.
Victoria Szabo: What are examples of actionable stories and calls to action? Can you define that? How do you measure success?
Frank Tramble: It is about what do we want people to do. We can’t do every single thing or idea. We want what moves an initiative forward. I expect to get calls on SCOTUS decisions, presidential cycle, DEI and other issues. How well do we navigate everything? Success is how well we navigate these things and how well we are intentional about it.
Tracy Futhey: Thank you for joining us today.
Tracy Futhey: While we are setting up for the next topic, I wanted to say that Matt Hirschey sent out a survey to faculty asking people about their experiences and interest with AI. We will have him speak at one of the Spring meetings to share what he has learned.
4:30 – 5:00pm - Virtual and Remote Computing and Virtual Apps for Students - David Lane, John Robinson (20-minutes presentation, 10 minutes Q&A)
What It Is: David Lane, John Robinson, and Mark McCahill will review the current offerings in virtual and remote computing platforms that enable student coursework. They will also discuss potential areas for growth in these service offerings.
Why It Is Relevant: Duke OIT provides and supports virtual computing infrastructures for Duke instructors/students which are often essential to research activities, homework & coursework. ITAC members will learn what is happening in this space and what changes can be expected in the near future.
Link to Presentation: "C:\Users\jbp60\Box\Information Technology Advisory Council\ITAC\Speaker Presentations\2023-24\020824 ITAC\Virtual and Remote Computing - ITAC.pptx"
John Robinson: We want to talk about virtual & remote computational things we do, along with some of the virtual apps that are available for students.
To recap, Virtual Computing Manager (VCM) and Container Manager (CM) have been in existence for a while and are supported by OIT. These services provide students, faculty and staff access to specialty apps that are not necessarily on a device. This is ideal for course work, projects, and development - something we would consider non-production but critical to coursework. At the beginning of the semester, we get a good understanding of what will need to be pre-provisioned, so that when students start class, they can reserve the resources at that time. Where we have challenges are when we don’t know how many students will be create reservations – this can cause congestion.
We are also reviewing our support model. We are currently working with OIT’s Device Engineering team (they create and package apps we support) with the goal of having continuity of the user-interface between the physical and virtual environments. We also want to increase the depth of knowledge about VCM and CM within this support team. This will increase the level of support provided to our users. .
Tracy Futhey: To clarify, Virtual Compute Manager and Container Manager are different from the words people sometimes use -- I need a virtual machine. We already have what is potentially a confusing landscape for faculty and students. Something we need to work on is how we explain the differences in theses virtual spaces and how are they used.
John Robinson: We could update our marketing efforts and communications material to better explain these services and potential use cases. The information on the slides we provided is for the most recent semester. We have also included a slide on historical trends. Container Manager is a mature offering that has been around for 8 years. We are pleased with the usage data and will continue to make use of it.
Dave Mc Alpine: Container Manager is fantastic for classes. Are you seeing a decline in usage now that Google Co-lab is making it easy to do in the cloud as well?
John Robinson: Based on what we are seeing there has been no real decline.
Charlie Kneifel: I think it’s the overhead of students setting it up for themselves. They can do it, but on that first day of class, it’s just another PC and now I’ve got to spend the first couple of classes helping you get your machine set up. Where in this case, you can choose to do it later yourself but when you start the class, you can use this.
John Robinson: Having it pre-provisioned is ideal.
Charlie Kneifel: There are cases where it’s not just the provisioning it’s the “class of three hundred students one, two, three click” things being done that can be challenging for the back-end infrastructure and the network.
Colin Rundel: We are happy users of the Container Managers and are working with OIT to customize for specific classes – it has been super helpful.
David McAlpine: I agree. Mark has helped us as well. We can build things out and push them in real time, which is great. But you can do the same in other programs. Is this a thing we need to keep up, since other areas offer the equivalent for free in Google?
Mark McCahill: One thing that’s different is that we can work in the back end to make the machine bigger if necessary. That isn’t as much an option in Google Colab.
Evan Levine: I don’t view them as competing services either. I would rather think of them as different pieces of a portfolio of services. When Tracy is talking about how we brand Virtual Management and how we make sense of it all, that actually should be another piece of it.
John Robinson: Another offering is Splashtop. This is now supported by OIT Device Engineering. Splashtop is a cloud offering to provide remote desktop into lab machines. One nice component is that it has a scheduling component to it. It lets students access computer labs anytime. Today, we have about three hundred users spread across one hundred and fifty machines. The lab administrators keep the schedule going.
I also want to mention Rapid VMs. It’s similar to the VCM environment, but more focused on research computing. And you can specialize for research needs.
John Board: How has Splashtop performed with graphically intensive applications?
Victoria Szabo: One of the reasons we really like Splashtop is because it uses the power of that machine. It works very well.
John Robinson: Part of the challenges we have heard with the other services has been the lack of graphically intensive GPU. It was brought up in the research needs conversations. Splashtop does fill that need (to a degree) however it is manually scheduled and there is no course integration. It does have limitations. Mark will talk about Bitfusion.
Mark McCahill: Bitfusion gives Linux machines access to shared GPUs. What’s good about this is that we can slice the GPUs up for multiple people to share. This can help support large classes. Duke Kushan University is also running Bitfusion clusters. The bad news is that Bitfusion ended availability to new customers as of May 5, 2023. They will keep supporting it until May 5, 2025. We will need to consider our options.
John Robinson: As we evaluate replacement options for Bitfusion, one are of focus will be Graphically Intensive GPUs.
David Lane: We have been in a yearlong proof of concept testing for VMware Horizon. Post corporate acquisition, they want to sell off their end user computing section. We think they will continue to exist as a business. VMware Horizon is a virtual machine manager, and it works very well. It’s a competitor to Citrix. In conjunction we are using an application from Nvidia, currently called Virtual Workstation. It takes a physical GPU and carves it into chunks and assigns it to a virtual machine. We can then assign it to as many as fifty to seventy users consuming one GPU. It is currently running on loaner hardware by several vendors. The loaner are provided for the duration of proof of concept. This is good, because the software isn’t inexpensive. We are continuing to get feedback from end users.
John Robinson: The proof of concept has been around for a year. We could have launched VMware Horizon earlier, however we wanted thorough testing and gather feedback. We reached out to Ernesto Escobar and his team – they provided valuable feedback. Other heavy users of Splashtop have been testing. So far, the feedback is positive. We will continue testing by other user groups.
We have discussed the idea of Nicholas School, instead of replacing their lab machines, replace them with monitors. The students still have the same experience. We also want to bring up Microsoft Azure Virtual Desktop (AVD) and Azure labs. We are investigating that and what their guardrails are.
Mark McCahill: Juicelabs.co is potential alternative to Bitfusion. It also supports Direct X. The company is a start-up. It’s a potential solution and works well, but there are bugs. We are helping them uncover some of the issues.
David Lane: Bitfusion was not a VMware product when we first started using them. Juice labs as a start- up may get acquired by someone else.
John Robinson: We’ve covered some of the challenges, but we need to mention others. Cost is always a factor, so is scalability and the longevity of the product. Currently we are making sure the product works the way we want it to. Next up would be planning on how we roll it out. With Azure we are looking at costs, how to control overutilization, how to do scheduling. What does it mean outside of class?
If there is down time, can we use the resources for other areas? We need to schedule appropriately.
David Lane: We are looking at two different licensing costs. There is a possibility that we just look at the NVidia software. With some development hours we might could use the Nvidia software with our existing VCM offering. We could take advantage of what already works well, but we haven’t run it by the automation team yet.
Victoria Szabo: Some software licenses don’t want you to be on a VM. How do we handle that aspect of it?
John Robinson: We have to work though that within the evaluation, to make sure we are in compliance. We need to push to make sure we communicate well as appropriate. We try to ensure that it is used properly.
Victoria Szabo: What if people ignore that?
David Lane: Guardrails have to be put in place. It wouldn’t be open for the entire student body, only for certain courses.
John Robinson: By using central resources, we have continuity. We would have checks and balances in place.
John Board: To add to your point, some of vendors want to audit our system for unauthorized usage.
Colin Rundel: Question regarding memory for the research toolkit model. Is there a potential for increasing the allocation?
Mark McCahill: The basic allocation for a student is that you can have one VM. In some courses the VMs can be larger. This is on a course-by-course basis.
Tracy Futhey: It is something that we offer, you just need to let us know.
Colin Rundel: It would be good to communicate it clearly on the website.
Matt Hirschey: By using remote desktops, how much money do we save by not having a student download software on their own computer?
Mark McCahill: Part of this is how much time the instructor saves.
Tracy Futhey: Evan can speak to this as a replacement for computer labs.
Evan Levine: Costs saving varies depending on which software. We had an issue with Maple Containers last night. Maple student licenses can be cheaper. It can add up, but supporting 300 students installs, laptops, etc. can be expensive to provide services.
John Robinson: Not every student machines can handle the applications at the same speed of level.
Ed Gomes: Part of the rationale when we first implemented it was equal access. It was also super critical during Covid, to make resources available. We feel we are getting our monies worth out of it.
Victoria Szabo: With things like Splashtop, there are lab licenses that can’t be used on a student machine.
Tracy Futhey: Yes, and machines in the labs are expensive.
Victoria Szabo: And it is good for faculty to have consistent environments for the students.
Ed Gomes: Faculty might expect certain plug-ins to be available.
Evan Levine: Some things we can package with the software.
Charlie Kneifel: This is good for updates and when not to update. We can choose not to upgrade.
John Board: Some of my TAs had good luck with having students install Docker on their own machines. And then run the containers locally on their machines.
Tracy Futhey: That’s another service and option to look at in terms of what use cases that would be useful.
Victoria Szabo: I’m surprised by the number of tablets coming in now.
Victoria Szabo: We are postponing that last discussion (below) for another date.
Meeting adjourned.
5:00 – 5:15pm - Updates from the Latest CSG meeting - Mark McCahill, Charley Kneifel (15 minutes)
What It Is: The Common Solutions Group (CSG) is comprised of a small set of research universities that participate regularly in meetings and project work. These CSG members are characterized by strategic technical vision, strong leadership, and have both ability & willingness to adopt common solutions on their campuses.
Why It Is Relevant: At CSG meetings, the members engage in detailed, interactive discussions of strategic technical and policy issues affecting research-university IT across time. We will get updates from the most recent meeting.
------------------------
[Topic deferred to future meeting]