DRAFT Minutes, ITANA Conference Call
June 15, 2007
 
Jim Phelps, University of Wisconsin-Madison (chair)
Gary Chapman, New York University
Hébert Díaz-Flores, University of California-Berkeley
Matt Kolb, Michigan State University
Tom Barton, University of Chicago
Renee Martin, North Carolina A&T
Steve Mullins, University of Alaska
Sue Sharpton, University of Alaska
Brendan Bellina, University of Southern California
Keith Hazelton, University of Wisconsin - Madison
Ann West, Educause/Internet2
Steve Olshansky, Internet2
Dean Woodbeck, Internet2 (scribe)
 
**Action Items**
 
Hébert will add information to the wiki in terms of a survey concerning Service Oriented Architecture.

All are invited to add items to the EA Facets wiki page.

Agenda, June 15, 200

Roll Call.

  • Agenda Bash.
  • Accept minutes of last call
  • Taxonomy of Pain and Carrots to offer (see EA Pain Points)
  • Enterprise Architecture as Strategy and EA Facets (see EA Facets)
  • SOA

Items on the shelf:

  • Mellon ESB Assessment - goal? is there date on this? Mark will try to get Chas DiFatta on a future call to talk about this.
  • EDUCAUSE Full-day seminar on Identity Management at EDUCAUSE 2007 (Request for case studies with different vendor solutions)
  • What do we do if we want a project done in the enterprise way and if you do, this would be good.

Carrots to offer

Hébert posted a note to the email list which kicked off a discussion about incentives, or carrots, that people use to reward enterprise-friendly activities. While central IT people are likely in the decision-making chain and able to see the benefits of taking into account the entire enterprise in making a decision, others may not.

A campus charge-back system, for example, may be a disincentive. Departments will make decisions based on time-to-deliver and cost. Thus, IT and purchasing decisions are made on a project-by-project basis and may not adequately consider the higher-level organizational goals.

An example is the consideration of content management systems for web publishing. Some of the data to be used will likely be enterprise core data, perhaps information about courses and instructors and departments. So if the institution purchases a common CMS, it will be easier to use that information in multiple places. A carrot may be to have the institution purchase the CMS centrally and offer it to individual units at no charge. Another carrot in this situation would be to provide someone from central IT to accomplish much of the work that will be involved.

One issue is to provide central funding for basic infrastructure. You can build the infrastructure and offer it to the campus just like you offer phone services and authentication. If you do something like this on a project-by-project basis, the early adopters will bear the brunt of the cost and time involved.

Gary Chapman said that NYU is planning to build an enterprise-wide identity management (IdM) system by building the infrastructure then doing a pilot project. The challenge for them isn't the funding, but that the staff resources are not there to do the build-out.

Tom Barton said that the University of Chicago has a project management office. That office sees all of the projects across campus and manages cost and deadlines. Once this office became convinced of the value of enterprise-wide decisions, that perspective became part of the decision-making process.

Jim Phelps reported that the University of Wisconsin has found that infrastructure can be built when there is enough distributed pain and people realize that building, say, an identity management infrastructure on their own is painful and is better done centrally.

Matt Cole reported that Michigan State is in the process of developing a lot of middleware to link existing applications together. One major driver has been a major restructuring of the HR and financial infrastructure at the university, which has now led to some enterprise-wide decisions in areas like identity management.

Brendan Bellina said that USC has a directory steering committee, which has become a clearinghouse for architecture decisions. Anyone with a project involving people needs to present a plan to this committee, which includes leaders from around the university, including IT leaders from departments and business leaders. Many times, someone will bring a concept forward and find that others are considering the same thing. Committee members will then recognize that the project should become a central IT function.

In general, then, transparency within the organization and between departments is essential. If campus offices and departments see and understand what IT is planning and doing, they are more likely to develop trust and enter a positive working relationship. One way of doing this is to draw people in to projects and also participate in any working groups or project in which you can help.

In summary, one incentive is to offer both the infrastructure and to fund and perform at least part of the implementation/integration work (as in the CMS example). Another is to demonstrate cost-saving across projects. If you have five projects and, in all five, you are paying to build the same thing, it may be more cost-effective to build that thing once. Disincentives include chargebacks and building out an infrastructure on a project-by-project basis.

Enterprise Architecture as Strategy and EA Facets

The book, Enterprise Architecture as Strategy (Ross, Weill, and Robertson), posits that there are certain things that should be core to the enterprise and certain things that need to be aligned. The question is how you identify these core items. A suggested first step is to determine where your organization fits on an operating model and maturity model.

The operating models run from very distributed to highly unified:

  • Distributed (few common business processes and little data integration)
  • Replicated (common business processes with little data integration)
  • Coordinated (few common business processes with high data integration)
  • Unified (common business process with high data integration)

Likewise, the maturity models are on a continuum:

  • Business Silos
  • Standardized Technology
  • Optimized Core
  • Business Modularity

Jim used the enrollment management operation at Wisconsin as an example. In terms of the operating model, enrollment management is pretty unified. Most undergraduates have unified data and the data is highly integrated. There is standardized technology. The maturity model is moving to the "shared services" step.

Another example at Wisconsin is the course management system. The university has one course management system (Desire2Learn), but other software is now popping up because the unified course management system is not meeting everyone's use case(s). That makes sense, given that instruction is highly diversified. So perhaps requiring one course management system was not the best business decision.

Hébert wondered, then if what is needed is a way to support multiple course management systems in some orderly way. NYU is using Sakai, an online collaboration and learning environment, hoping that it may provide some of the necessary flexibility.

Jim suggested fashioning a survey based on EA facets and using the information to help organizations determine their maturity models. This might be used to allow contact with those, say, a step ahead of your organization on the model.

The EA Facets page of the wiki includes a list of facets that could form the basis for a survey. All are invited to add items to this wiki page

Service Oriented Architecture

There was discussion about developing a survey to gather information and metrics concerning how institutions are implementing SOA. Hébertwill add information to the wiki in this regard.
 
Next Call is July 13, 2007