BlogU

  • 2020: The Campus Data Center Follows the Computer Lab

    By Joshua Kim July 22, 2010 9:13 pm EDT

    This prediction can't be correct. If the campus data center is really going to suffer the same fate as the computer lab by 2020 then something is really strange with our current reality. We hear everyday about how the data center is growing faster than we keep up. Not enough power. Not enough cooling. Not enough storage. We are virtualizing, expanding and collocating. It is never enough.

     

    But couldn't we have said the same thing about the campus computer lab in 2000? Weren't we facing a crisis of over-subscribed computer labs back then? Wasn't much of the talk about replacement cycles, and the potential of thin clients to alleviate the demands of buying full-blown computers for our labs? Did we really imagine that for many of us (I know, some of us are still building computer labs today), that the availability of inexpensive laptops and netbooks would be driving the computer lab into obsolescence?

     

    What would it take for our data center's to migrate to the cloud by 2020?:

     

    Bandwidth: In 2010 I'm still doubtful that hosting all of our campus media in the cloud is feasible. As the amount of video on our networks increase (from lecture capture, rapid authoring, and streaming of curricular media), it is unclear to me what would occur if all of this video (increasing in hi-def) would perform if it all had to be uploaded and retrieved from the cloud. When will our network gateways be ready for this type of traffic?

     

    Trust: We don't seem ready to trust our data to other people. I'm not talking about FERPA protected data, which I'm not sure will ever be stored outside of our campus owned systems (although I could be wrong). I'm thinking of even our digital curricular and library content. Teaching and learning depends on our LMS and library systems being available at all times - and most of us are nervous about the idea of not controlling this content.

     

    Risk: Who is going to make the first decision to move all their campus data to the cloud? This seems like it could be a career ending decision if things go badly. While many services seem to be moving to the cloud (such as e-mail and calendaring), it still appears that the demands for servers and storage are growing at a faster rate than any cloud migration (hence the move to virtualization etc.). Someone is going to have to be willing to leapfrog current practices. Someone is going to have to take a risk. Is this being done anywhere? Do we have examples of an entire campus that consumes all of their data center applications and storage needs as a service?

     

    What is your prediction for the year 2020?

Go to comments (2) »
Advertisement
Don't Miss a Great Opportunity

Comments on 2020: The Campus Data Center Follows the Computer Lab

  • what does " to migrate to the cloud" mean?
  • Posted by Henry , OIT at NC State Univ on July 25, 2010 at 8:30pm EDT

  • "the cloud" is misleading - there isn't only one - there are many clouds, public and private. NC State runs a private cloud http://vcl.ncsu.edu/ for academic use including both homework and research.

    The rest of the argument is so general as to lose meaning. Of course any competent ICT administration will evaluate TCO, and users will look at cost/benefit. Of course the "people" issues are important - and many people are "server huggers" - my prediction is that most of this is unjustified and will decrease as costs are carefully scrutinized.

  • Small to mid-size data centers are like Studebakers
  • Posted by Dirk Herr-Hoyman , eLearning System Architect on July 30, 2010 at 4:45pm EDT

  • I think that small to mid-sized data centers are on their way out. This is all over the board, not just in higher ed. You will soon, if not already, be able to buy the same services, which are sometimes called SaaS, at a lower cost and better security than you can get in a small to mid-sized data center. It's about volume, what you can do at scale in a data center. I was at a presentation by Brian Hawkins, then President of Educause, circa 2001, where he said much the same. I think Brian was right now, though at the time I was mildly skeptical. This is what I will term the Internet 2.0, with apologies to my many colleagues/friends who are working for the I2 project.

    It's not hard to predict this happening, all the trends of networks, hardware, and applications are pointing in this direction. And the same sort of thing has happened in many other technologies, like the auto industry. I grew up about 25 mi from the main Studebaker plant and watched it collapse taking the city of South Bend, IN with it for a while. In a generation we went from 1000s of auto makers to 10s. Which was about volume and what you can do at scale.

    When this does happen, there will be a flip-flop in attitudes about the so-called cloud computing. One day it will be a villian, the next a hero. One day it won't be up to our FERPA standards, the next it will be the gold standard. That's the "tipping point" in action, when we hit the inflection point on the S-curve of the difusion of innovation. I do think that costs are what will be the driver with the so-called commodity applications/technologies, like email, being thrd e first to flip. Data storage is coming, with a bunch of low cost internet storage providers cropping up all over the place taking advantage of the next generation of solid state storage (we don't need no stinkin' hard drives) Network bandwidth may be a limiting factor at the moment, but I see Moore's Law (at least) happening there in a big way. Backup, or challenges in getting a good one, is another fundamental factor which will drive us towards cloud-based storage. That's the dirty secret in IT, it's getting harder and harded to keep up you backups, what with explosion of disk space. A whole different strategy, one that's more like holographic storage, where everything is storage in a highly redundant manner that makes a backup irrelevant, that's where we are headed.

    When will this really hit and wipe out the small to mid-sized data centers? 10-20 years is what it looks like to me. It's already beginning to happen, so I may be conservative on saying 20 years.
    Oddly enough, I do see a place for computer labs and niche data centers, ones that are small and somewhat specialized, in the higher ed landscape. But the traditional campus data center, well that will only be on a large campus, the R1s of the world. And they will be a hub for many other campus data centers, which they will be colocating. Which is already starting to happen.

    This isn't going to put folks out of work on those small to mid-sized campuses. No, not really. They will just get different IT things to do. Just like the folks who used to do mainframe, VAX/VMS, DOS, WANG (and I'm sure there's a few more :-) all found new things to do.

Print
© Copyright 2011 Inside Higher Ed