The next BriefingsDirect Voice of the Customer IT infrastructure thought leadership case study explores how Purdue University has created a strategic IT environment to support dynamic workload requirements.
We'll now hear how Purdue extended a research and development IT support infrastructure to provide a common and "operational credibility" approach to support myriad types of compute demands by end users and departments.
To describe how a public university is moving toward IT as a service, please join Gerry McCartney, Chief Information Officer at Purdue University in Indiana. The discussion is moderated by BriefingsDirect's Dana Gardner, Principal Analyst at Interarbor Solutions.
Here are some excerpts:
Gardner: When you're in the business of IT infrastructure, you need to predict the future. How do you close the gap between what you think will be demanded of your infrastructure in a few years and what you need to put in place now?
McCartney: A lot of the job that we do is based on trust and people believing that we can be responsive to situations. The most effective way to show that right now is to respond to people’s issues today. If you can do that effectively, then you can present a case that you can take a forward-looking perspective and satisfy what you and they anticipate to be their needs.
I don’t think you can make forward-looking statements credibly, especially to a somewhat cynical group of users, if you're not able to satisfy today’s needs. We refer to that as operational credibility. I don’t like the term operational excellence, but are you credible in what you provide? Do people believe you when you speak?
Gardner: We hear an awful lot about digital disruption in other industries. We see big examples of it in taxi cabs, for example, or hospitality. Is there digital disruption going on at university campuses as well, and how would you describe that?
McCartney: A university you can think of as consisting of three main lines of business, two of which are our core activities, of teaching students, educating students; and then producing new knowledge or doing research. The third is the business of running that business, and how do you do that. A very large infrastructure is built up around that third leg, for a variety of reasons.
But if we look at the first two, research in particular, which is where we started, this concept of the third leg of science has been around for some time now. It used to be just experimentation and theory creations. You create a theory, then you do an experiment with some test tubes or something like this, or grow a crop in the field. Then, you would refine your theory and you would continue in that kind of dyadic mode of just going backward and forward.
Third leg of science
That was all right until we wanted to crash lorries into walls or to fly a probe into the sun. You don’t get to do that a thousand times, because you can’t afford it, or it’s too big or too small. Simulation has now become what we refer to as the third leg of science.
Slightly more than 35 percent of our actual research now uses high-performance computing (HPC) in some key parts of it to produce results, then shape the theory formulation, and the actual experimentation, which obviously still goes on.
Around teaching, we've seen for-profit universities, and we've seen massive open online courses (MOOCs) more recently. There's a strong sense that the current mode of instructional delivery cannot stay the same as it has been for the last hundreds of years and that it’s ripe for reform.
Indeed, my boss at Purdue, Mitch Daniels, would be a clear and vibrant voice in that debate himself. To go back to my earlier comments, our job there is to be able to provide credible alternatives, credible solutions to ideas as they emerge. We still haven’t figured that out collectively as an industry, but that’s something that is in the forefront of a lot of peoples’ minds.
Gardner: Suffice to say that information technology will play a major role in that, whatever it is.
McCartney: It’s hard to imagine a solution that isn’t actually completely dependent upon information technology, for at least its delivery, and maybe for more than that.
Gardner: So, high-performance computing is a bedrock for the simulations needed in modern research. Has that provided you with a good stepping stone toward more cloud-based, distributed computing-based fabric, and ultimately composable infrastructure-based environments?
McCartney: Indeed it has. I can go back maybe seven or eight years at our place, and we had close to 70 data centers on our campus. And by a data center, I mean a room with at least 200-amp supply, and at least 30 tons of additional cooling, not just a room that happens to have some computers in it. I couldn't possibly count how many of them there are now. Those stand-alone data centers are almost all gone now, thanks to our community cluster program, and the long game is that we probably won't have much hardware on our campus at some point a few years from now.
Right now, our principal requirement is around research computing, because we have to put the storage close to the compute. That's just a requirement of the technology.
In fact, many of our administrative services right now are provided by cloud providers. Our users are completely oblivious to that, but we have no on-premises solution at all. We're not doing travel, expense reimbursement and a variety of back-office things on our campus at all.
That trend is going to continue, and the forcing function there is that I can't spend enough on security to protect all the assets I have. So, rather than spend even more on security and fail to provide that completely secure environment, it's better to go to somebody who can provide that environment.
Gardner: What sort of an infrastructure software environment do you think will give you that opportunity to make the right choices when you decide on-prem versus cloud, even for those intensive workloads that require a tight data and compute link?
McCartney: The worry for any CIO is that the only thing I have that's mine is my business data. Anything else -- web services, network services -- I can buy from a vendor. What nobody else can provide me are my actual accounts, if you wish to just choose a business term, but that can be research information, instructional information, or just regular bookkeeping information.
When you come into a room of a new solution, you're immediately looking at the exit door. In other words, when I have to leave, how easy, difficult, or expensive is it going to be to extract my information back from the solution?
That drives a huge part of any consideration, whether it's cloud or on-prem or whether it's proprietary or open code solution. When this product dies, the company goes bust, we lose interest in it, or whatever -- how easy, expensive, difficult is it for me to extract my business data back from that environment, because I am going to need to do that?
Gardner: What, at this juncture, meets that requirement in your mind? We've heard a lot recently about container technology, standards for open-source platforms, industry accepted norms for cloud platforms. What do you think reduces your risk at this point?
McCartney: I don't think it's there yet for me. I'm happy to have, relatively speaking, small lines of business. Also, you're dependent then on your network availability and volume. So, I'm quite happy there, because I wasn't the first, and because that's not an important narrative for us as an institution.
I'm quite happy for everybody else to knock the bumps out of the road for me, and I'll be happy to drive along it when it’s a six-lane highway. Right now it's barely paved, and I'll allow other brave souls to go there ahead of me.
Gardner: You mentioned early on in our discussion the word "cynical." Tell me a little bit about the unique requirements in a university environment where you need to provide a common, centrally managed approach to IT for cost and security and manageability, but also see to the unique concerns and requirements of individual stakeholders?
McCartney: All universities are, as they should be, full of self-consciously very smart people who are all convinced they could do a job, any particular job, better than the incumbent is doing it. Having said that, the vast bulk of them have very little interest in anything to do with infrastructure.
The way this plays out is that the central IT group provides the core base that services the network -- the wireless services, base storage, base compute, things like that. As you move to the edge, the things that make a difference at the edge.
Providing the service
In other words, if you have a unique electrical device that you want to plug in to a socket in the wall because you are in paleontology, cell biology, or organic chemistry, that's fine. You don't need your own electricity generating plants to do that. I can provide you with the electricity. You just need the cute device and you can do your business, and everybody is happy.
Whatever the IT equivalent to that is, I want to be the energy supplier. Then, you have your device at the edge that makes a difference for you. You don't have to worry about the electricity working; it's just there. I go back to that phrase "operational credibility." Are we genuinely surprised when the service doesn’t work? That’s what credibility means.
Gardner: So, to me, that really starts to mean IT as a service, not just electricity or compute or storage. It's really the function of IT. Is that in line with your thinking, and how would you best describe IT as a service?
McCartney: I think that's exactly right, Dana. There are two components to this. There's an operational component, which is, are you a credible provider of whatever the institution decides the services are that it needs, lighting, air-conditioning or the IT equivalence of that? They just work. They work at reasonable cost; it's all good. That’s the operational component.
The difference with IT, as opposed to other infrastructure components, is that IT has itself the capability to transform entire processes. That’s not true of other infrastructure things. I can take an IT process and completely reengineer something that's important to me, using advantages that the technology gives me.
For example, I might be concerned about student performance in particular programs. I can use geo-location data about their movement. I can use network activity. I can use a variety of other resources available to me to help in the guidance of those students on what’s good behavior and what’s helpful behavior to an outcome that they want. You can’t do that with an air-conditioning system.
IT has that capability to reinvent itself and reinvent entire processes. You mentioned some of them the way that things like Uber has entirely disrupted the taxi industry. I’d say the same thing here.
There's one part of the CIO’s job that’s operational; does everything work? The second part is, if we're in transition period to a new business model, how involved are the IT leaders in your group in that discussion? It's not just can we do this with IT or not, but it’s more can a CIO and the CIO’s staff bring an imagination to the conversation, that is a different perspective than other voices in the organization? That's true of any industry or line of business.
Are you merely there as a handmaiden waiting to be told what to do, or are you an active partner in the conversation? Are you a business partner? I know that’s a phrase people like to use. There's a kind of a great divide there.
Gardner: I can see where IT is a disruptor -- and it’s also a solution to the disruptor, but that solution might further disrupt things. So, it's really an interesting period. Tell me a little bit more about this concept of student retention using new technologies -- geolocation for example -- as well as big data which has become more available at much lower cost. You might even think of analytics as a service as another component of IT as a service.
How impactful will that be on how you can manage your campus, not only for student retention, but perhaps for other aspects of a smarter intelligent campus opportunity? [See related post, Nottingham Trent University Elevates Big Data’s Role to Improving Student Retention in Higher Education.]
McCartney: One of the great attractions of small educational institutions is that you get a lot of personalized attention. The constraint of a small institution is that you have very little choice. There's a small number of faculty, and they simply can’t offer the options and different concentrations that you get in a large institution.
In a large institution, you have the exact opposite problem. You have many, many choices, perhaps even too many subjects that, as a 19-year-old, you've never even heard of. Perhaps you get less individualized attention and you fill that gap by taking advice from students who went to your high school a year before, who are people in your residence hall, or people you bump into on the street. The knowledge that you acquire there is accidental, opportunistic, and not structured in any way around you as an individual, but it’s better than nothing.
There are advisors, of course, and there are people, but you don't know these individuals. You have to go and form relationships with them and they have to understand you and you have to understand them.
A big-data opportunity here is to be able to look at the students at some level of individuality. "Look, this is your past, this is what you have done, this is what you think, and this is the behavior that we are not sure you're engaging in right now. Have you thought about this path, have you thought about this kind of behavior for yourself?"
A well-established principle in student services is that the best indicator of student success is how engaged they are in the institution. There are many surrogate measures of that, like whether they participate in clubs. Do they go home every weekend, indicating they are not really engaged, that they haven’t made that transition?
Independent of your academic ability, your SAT scores, and your GPA that you got in high school, for students that engage, that behavior is highly correlated with success and good outcomes, the outcomes everybody wants.
As an institution, how do you advise or counsel. They'll say perhaps there's nothing here they're interested in, and that can be a problem with a small institution. It's very intimate. Everybody says, "Dana, we can see you're not having a great time. Would you like to join the chess club or the drafts club?" And you say, "Well, I was looking for the Legion of Doom Club, and you don’t seem to have one here."
Well, you go to a large institution, they probably have two of those things, but how would you find it and how would you even know to look for that? How would you discover new things that you didn't even know you liked, because the high school you went to didn't teach applied engineering or a whole pile of other things, for that matter.
Gardner: It’s interesting when you look at it that way. The student retention equation is, in a business sense, the equivalent of user experience, personalization, engagement, share of wallet, those sorts of metrics.
We have the opportunity now, probably for the first time, to use big data, Internet of Things (IoT), and analytics to measure, predict, and intercede at a behavioral level. So in this case, to make somebody a productive member of society at a capacity they might miss and you only have one or two chances at that, seems like a rather monumental opportunity.
McCartney: You’re exactly right, Dana. I'm not sure I like the equivalence with a customer, but I get the point that you're making there. What you're trying to do is to genuinely help students discover an effective path for themselves and learn that. You can learn it randomly, and that's nice. We don't want to create this kind of railroad track. Well, you're here; you’ve got to end up over there. That’s not helpful either.
My own experience, and I don’t know about other people listening to this, is that you have remarkably little information when you're making these choices at 19 and 20. Usually, if you were getting direction, it was from somebody who had a plan for you that was more based on their experience of life, some 20 or 30 years previously than on your experience of life.
So where big data can be a very effective play here, was to say, "Look, here are people that look like you, and here were the choices they've made. You might find some of these choices interesting. If you might, then here’s how you’d go about exploring that."
As you rightly say, and implicitly suggested, there is a concern with the high costs, especially of residential education, right now. The most wasteful expenditures there are is where you do a year or two to find out you shouldn't have ever been in this program, you have no love for this thing, you have no affinity for it.
The sooner you can find that out for yourself and make a conscious choice the better. We see big data having a very active role in that because one of the great advantages of being in a large institution is that we have tens of thousands of students over many years. We know what those outcomes look like, and we know different choices that different people have made. Yes, you can be the first person to make a brand new choice, and good for you if you are.
Gardner: Well it’s an interesting way of looking at big data that has a major societal benefit in the offing. It also provides predictability and tools for people in ways they hadn’t had before. So, I think it’s very commendable.
Before we sign-off, what comes next – high performance computing (HPC), fabric cloud, IT-as-a service -- is there another chapter on this journey that perhaps you have a bead on that that we’re not aware of?
McCartney: Oh my goodness, yes. We have an event now that I started three years ago called "Dawn or Doom," in which if technology is a forcing function, if it is. We're not even going to assert that definitely. Are we reaching a point of a new nirvana, a new human paradise where we’ve resolved all major social problems, and health problems or have we created some new seventh circle of hell where it’s actually an unmitigated disaster for almost everybody; if not everybody? Is this the end of life as we know it? We create robots that are superior to us in every way and we become just some intermediate form of life that has reached the end of its cycle.
This is an annual event that's free and open. Anybody who wants to come is very welcome to attend. You can Google "Dawn or Doom Purdue." We look at it from all different perspectives. So, we have obviously engineers and computer scientists, but we have psychologists, we have labor economists. What about the future of work? If nobody has a job, is that a blessing or a curse?
Psychologists, philosophers, what does it mean, what does artificial intelligence mean, what does a self-conscious machine mean? Currently, of course, we have things like food security we worry about. And the Zika virus -- are we spawning a whole new set of viruses we have no cure for? Have we reached the end of the effectiveness of antibiotics or not?
These are all incredibly interesting questions I would think any intelligent person would want to at least probe around, and we've had some significant success with that.
Gardner: When is the next Dawn or Doom event, and where will it be?
McCartney: It would be in West Lafayette, Indiana, on October 3 and 4. We have a number of external high-profile key note speakers, then we have a passel of Purdue faculty. So, you will find something that entertain even the most arcane of interests. [For more on Dawn or Doom, see the book, Dawn or Doom: The Risks and Rewards of Emerging Technologies.]
You may also be interested in:
- Securing data provides Canadian online bank rapid path to new credit card business
- Nottingham Trent University Elevates Big Data’s Role to Improving Student Retention in Higher Education
- How Allegiant Air solved its PCI problem and got a whole lot better security culture, too
- How European GDPR compliance enables enterprises to both gain data privacy and improve their bottom lines
- DevOps and security, a match made in heaven
- Alation centralizes data knowledge by employing machine learning and crowdsourcing
- Expert panel explores the new reality for cloud security and trusted mobile apps delivery
- Catbird CTO on why new security models are essential for highly virtualized data centers
- Intralinks Uses Hybrid Cloud to Blaze a Compliance Trail Across the Regulatory Minefield of Data Soveriegnty
- 451 analyst Berkholz on how DevOps, automation and orchestration combine for continuous apps delivery
- Business unusual: How the Dell-EMC merger sends shockwaves across the global storage market
- Playtika bets on big data analytics to deliver captivating social gaming experiences
- Capgemini and HPE Team Up to Foster Behavioral Change That Brings Better Cyber Security Across Application Lifecycles