Tuesday, April 3, 2012

What is a platform?


As you may know, the Unit IT Steering Committee is helping to create a Strategic Technology Plan for the university. Next week, we're having a 2-day retreat to begin to work out some of the details.


Within the committee, there has been a lot of discussion regarding taking a “platform approach.”  At next week’s retreat, much of the planned discussion is around these platforms.  And yet, I don’t have a well-defined understanding of what people mean when they say “platform.”  Is it role based (a student platform; a researcher platform)?  Is it knowledge-based (a platform for creating data; a platform for sharing data)?  Is it function-based (a platform for high-speed computing; a platform for collaboration)?  Is it something else completely?


Once we agree on the types of platforms, can we use the standard definition of a technology platform: "infrastructure upon which other technologies and applications can be built?" If so, where do we start? Once we agree on a classification from the previous paragraph, would we then start enumerating the hardware and/or software elements that would make up the platform, or is there another path we should follow?


In order for us to have the most productive discussion next week, I thought it would help to reach out to a broad audience to hear thoughts on strategic technology platforms. It would really help if you could add your comments to this post by Friday so we can see if there's a consensus around what constitutes a platform and how to go about defining the ones we should be building at Michigan.

13 comments:

  1. I agree - the important question is, "What kind of platform?" Another way of saying this is: "What will the platform help us do more easily? This helps us frame this discussion with respect to the University's core mission and strategic plans.

    I need to think on this more...
    Might be helpful to have a little more disambiguation. There are a couple of flavors of platform "technology platform"(1) and "computing platform"(2). It seems that the former is more enabling of a specific process and the later enables computing processes or development. Is it the former definition we mean, as in a teaching platform or research platform? ... or the latter... or neither? Wikipedia gives an easily understood example of a technology platform for the pharmaceutical industry.(1)

    - Elle

    (1) http://en.wikipedia.org/wiki/Platform_technology(2) http://en.wikipedia.org/wiki/Computing_platform

    ReplyDelete
  2. An existing platform that UM supports is the hardware architecture and application frameworks that CTools runs atop. UM has developers (plus a community of Sakai developers at other institutions) that write tools against this platform w/o being burdened with the intricacies and management of the underlying architecture. The Open Academic Environment (OAE) is an updated version of this platform that will allow an even greater abstraction from the underlying infrastructure and lower the barrier to creating new tools for teaching & learning.

    ReplyDelete
  3. My basis for thinking about this is that technology is subservient to the various functions of the university. It is an enabler. In that sense platform for me means a set of technologies to support these functions.
    Thus, there is a "platform for teaching", which may include technologies such us LMS, classroom AV, e-portfolios, online collaborative spaces, etc. (Platform for online learning might be somewhat different, including things such as web- or videoconferencing).
    There is a "platform for high-performance computing in research", which may include computing clusters, massive data storage, very high bandwidth, etc.
    There is a "core platform" that is supporting the work of UM faculty and staff at the fundamental level: e-mail, calendaring, file storage...
    There is a "collaboration platform", "administrative platform", etc. Once we know which technologies constitute a platform, we can then pick and choose specific tools, which are the most fluid element in any platform.

    ReplyDelete
    Replies
    1. I struggle with the potential use of the term "platform." In many cases above, I see the list as a set of services supporting a specific domain. Some of the services may include a platform but some will be services any faculty member, staff, or student can simply "use". A platform implies an environment which I can leverage to create something "on top of" to meet a need that meets a certain pattern.

      Delete
  4. This comment has been removed by the author.

    ReplyDelete
  5. Other examples of existing platforms are Purdue's HUBzero for research and their learning & teaching applications Signals, Mixable & Hotseat. I think these are very interesting because they are platforms for researchers, educators and students to use and not only computing platforms that enable the development of applications by skilled programmers. I think sometimes we as technologists are very keen on building computing platforms and less keen on making the application platforms on top of that technology. (No offense to Chris, who really is charged with making our technology platform - which is very important.)

    See http://campustechnology.com/Articles/2011/08/04/Reengineering-IT-in-Higher-Education.aspx?Page=1

    ReplyDelete
  6. My only comment really related to the standard definition: "infrastructure upon which other technologies and applications can be built?" is that the University really *must* make sure everything works now -- and in the future -- on as many "open technologies" as possible.

    Any example of something that fails this which was *just released centrally* -- is DART (which only runs on Windows -- using Internet Explorer). When something like this slides through the vetting process, it makes the whole concept of "centrally provided services" completely suspect. And that's really unfortunate.

    ReplyDelete
  7. With regard to DART, that project did not slide through a vetting process. As we know, there is no perfect technical solution (yet). In the decision to select the product to support the "development" community, the client browser support issue was known and the tradeoff with other factors, against other solutions, was made by our Executive Officers. As always, we push our vendors for more support as they enhance their offering.

    ReplyDelete
  8. You may be familiar with the expression "separation of mechanism and policy." Decomposing or unbundling complex or highly integrated systems into component parts may help show platforms. As the policy diminishes and the mechanism proportionally increases, I think some platform starts to emerge. What interfaces or points of access would be available into the components of, e.g., CTools, ITAM, and EUS/EUC besides the use of the fully integrated services? What components are shared across CTools, ITAM, and EUS/EUC?

    ReplyDelete
  9. I think one reason this question is so hard to answer is because the more useful platforms tend to have structures that hold technology to very specific requirements at deeper levels and in exchange can accommodate unexpected diversity at higher levels of functionality and still deliver an efficient architecture that results in a consistent end result.

    The Internet as a platform has built on the TCP-IP protocol which is very specific about what is permitted in the link layer, gives you some limited options for routing and prioritizing packets of data, and at the highest level, quite a bit of freedom about what tools you use to create the user experience at the applications level

    Maybe there is a discipline we can take from the OSI model, of identifying the levels of functionality where diversity of technology is least useful/least efficient, and levels where the user experience is paramount, and find the fewest levels of function to connect the two.

    Sometimes I think it is correct to let the use case drive the platform decision down to a certain level. Maybe a lot of the time. We aren't starting from scratch. We have a lot of platform in place.

    Also, I think we need to consider the expert support (people) that enable our users(faculty, students and staff) to be productive, as a layer in the model of what makes an efficient platform.

    ReplyDelete
  10. As I review the preparation materials for the retreat, I'm wondering what, if anything we can mine from our CTools (or even our SIS) platform to assess the teaching and learning that is going on using this and other tools. One of the suggestions of the Teaching and Learning domain was that UM could become a leader in learning assessment. I'm struggling with Phil Hanlon's comment that the complexity of our environment indicates that recognition of emergence is a more appropriate strategy than an outcome based plan. He said, "This institution is probably one of the most complex eco-systems, intellectual eco-systems, in contemporary society.... It is much more important to identify process than plan. Strategy is probably less important than the processes that you put in place to identify what is going on, to harvest good ideas, to allow them to propagate and so forth." You don't have to have a platform that proscribes what tools will be used, the platform could provide a framework in which various tools could be used while the platform itself provides an environment for assessment. Difficult to describe an example; this is somewhat theoretical thinking.

    ReplyDelete
  11. This is all pretty abstract which is to be expected because the focus is at a very high level. I like John W's comment re the Internet as a platform. That resonates but is too broad for our purposes? At the same time it seems that while Ctools may be a platform it seems too constricted for a university wide architecture. Is there something in between that can span the breadth and still give succor to the development of a large variety of sub-platforms? Am I anywhere close to asking a reasonable question?

    ReplyDelete
  12. I think Chris will come up with a good set of definitions and a description of a platform, services and portfolio. just to capture some extras I think I'm hearing from the discussion:
    - One descriptor of a platform is the capability for horizontal extensibility as drawn on the board by Paul.
    - There may be certain "enabling characteristics" of a UM platform such as a define protocol?/ architecture?/&-or / data structure? to allow implementation of standard services for such things as accounting, analysis, "business intelligence"
    - Perhaps a function of the service that creates the specific instantiation of a platform for a project or research effort could be the ability to identify opportunities for participation, e.g. graduate students, specific expertise needed, professional and student employees needed.

    ReplyDelete