Architecture Management Meeting

Date: 6 August 2009
Time: 7:00 AM Pacific, 10:00 AM Eastern, 3:00 PM UK, 4:00 PM Frankfurt, 5:00 PM Haifa
Call In Number: (emailed)
Participation request: contact JimConallen

Agenda

1. Brief introduction of new participants.
2. Discuss timeframe for milestones and goals.
3. Review scenarios

Minutes

Atendees: Jim Conallen, Maneesh Goyal, Vishy Ramaswamy, Marnie Andrews, Jin Li, Steve Abrams, Srinivasan Renganatha, Brenda Ellis, Jonathan Harclerode, Derry Davis, Eldad Palachi

1. Introduced Brenda and Jonathan
2. Everyone is to add their name to the participants list, and thereby verify ability to edit wiki. If have problems email/call me and I can help work through.
3. Milestones
   1. We need to establish a concrete milestone to start working towards. Even though we still are in early stages of understanding the key scenarios, I think we need to have a concrete target to look towards. So we can use this as a start (and refine as necessary)
   2. Draft specification due 26-Nov-2009 for a meta-data model, and REST interface to access AM resources (i.e. models) and its meta-data.
4. Steve pointed out that ideally these milestones should coordinate with shipping products. The only product reps we have on call are RRC and Rational’s Modeling on Jazz product (no official name yet). It would be nice if these products could identify expected releases for us to coordinate our milestones with.
5. Jim mentioned that OpenUP description distilled.
6. Ren discussed the process he is using that drives BPM into PIM and then PSM.
   1. Process, and information models drive PIM.
   2. Traceability is key
   3. Support impact analysis
7. Brenda mentioned that the mappings between PIM and PSM are important but that we really need links all the way up to requirements and to actual source (versions).
8. Ren agreed that links should connect specific versions of resources/artifacts.
9. Brenda went on to say that this fits into enterprise architecture, and that having links to baselined resource is important, especially when referencing defined enterprise architectures
10. Steve asked, “What do we need in our links to support these scenarios”
11. Before Brenda had to leave (early) she indicated that we need to consider product lines, and baselines. There are many enterprises that have enterprise models, or baselined artifacts that customized to develop as specific application.
12. Steve said the interesting discussions should be on how we use the links in the scenarios, for example with notification and impact analysis. How can we create effective queries to navigate the links. When artifacts change we might consider changing a dirty bit, or suspect bit (i.e. ReqPro? ) on the link to indicate that some analysis needs to be done to re-assert the integrity of the link. We can also support impact analysis via work items, and attaching the artifacts to it and using work items to navigate potential impacts.
13. Jonathan said that his team is already supporting impact analysis with tooling. The DSLs that he created and is using supports (to some degree) the linkages necessary to allow automation determine sets of models that are subject to impact given a change. They use a combination of proxies and links between data models and business process models, and requirements. They are using ReqPro? and RSA as primary requirements and modeling tools.
14. Johathan also stated that in his environment a Requirement is a requirement and a model is a model, and the two are managed separately. This clean separation of concerns helps them manage it all. They can provide answers to questions like “What requirement are satisfied by this business service”
15. Vishy contributed that links can/should have key properties associated with that individual customers can define and use.
16. Vishy also agreed that links should be between specific versions of artifacts/resources.
17. Steve mentioned that we can’t make too many assumptions about the ability the implementers of the specifications we are working on when it comes to accessing older versions. We will need to specify out what we need in the spec and let individual implementations provide a working solution.

Homework

• Ren and Derry are to update the wiki with the usage scenarios
• Vishy volunteered to start a definitions page which will contain brief definitions of the terms we are using.
• Jonathan will document an outline of the MDA/MDD scenario that he has been sucessful with, and most importantly a description of how it supports impact analysis

-- JimConallen - 06 Aug 2009