[CSG Fall 2008] Non-Enterprise Initiatives of SOA

Services that are reliable, specific, small, secure, stable, standards-based and reusable.

SOA slated at Stanford -

Remedy - Stanford uses it heavily in the central IT office for order fulfillment and billing, and tracking some labor hours. They want to expose it as a service to replace brittle current integrations. Will allow new functionality. They're building a Remedy-Zimbra integration

Tom Barton is talking about SOA at Chicago.

No bigg app yet, but the want to be prepared. They thingk web services seems like a good approach to solving some problems. Their early experience is that even with a single simple service, there are lots of issues.

WS & ID management - Identifier translation - translate among ChicagoID, SSN, NEtID, hospital ID, alumID, studentID, etc. Also did an account management integration project with the hospital.

Net of early experience - concentrate on functionality, not underlying data stores. Loose coupling allows nice divisions of labor.

Data services - might provide good technology for providing data across the institution. A combination of reporting & WS capabilities.

Steps toward first SOA - look for cowpaths in institutional data - a relatively small set of objects that reflect the way we think about our businesses. Choose web services to reinforce a common data model across systems. Facing infrastructure choices, like servlet platform, esb, etc and development framework.

Designing Grouper's WS Interfaces -

It takes 4 WS interface styles to please enough adopters - SOAP vs. REST, and heavy capabilities vs. lightweight ones.