Re: [soa-rm-ra] resend: thoughts on management section (2)

[Rex Brooks <rexb@starbourne.com>]

Thanks for some interesting and thought-provoking issues, Ken,

As seems to happen with regularity, you bring up an issue, in this 
case simulation, just as I (and others I'm working with) are actually 
working on this. It just so happens that NIST's manufacturing 
division has a Modeling and Simulation group with whom we're working 
on some tests using their exercise control system (for emergency 
management scenarios). I can see where the capability could be used 
to test cross-boundary service lifecycles in specific SOA Ecosystems.

Of course, this would be particularly appropriate for EGov SOA, and, 
perhaps we could aim at September - October demonstrations? The next 
SOA for  EGov Conference is scheduled for September 6 and I'm working 
on a Collaborative Expedition Workshop for October 14, and like the 
back to back presentations last week, these could be efficiently 
combined. It might be worth including our RA work on this, though 
right now may not be the best time to discuss it. Apparently my 
inner-Murphy was displeased with me for pulling off the back to back 
presentation sessions so I went out and had myself a nice little 
bicycle accident last Friday. Relatively minor injuries, but 
definitely continues to get my attention.

Sheesh!
Rex


At 8:48 AM -0400 5/7/08, Ken Laskey wrote:
>*** I realized the formatting got hosed and the first send of this 
>email may be difficult to decipher ***
>
>This is a follow-on to ramblings from Monday's email.
>
>I'm not sure to what extent this fits into areas covered by 
>management or where else it goes.
>
>development support
>* if new solutions in a SOA ecosystem are to make use of deployed 
>resources, what modes of interaction do deployed resources need to 
>support?
>* lot of talk about needing to simulate performance - to what extent 
>would we like each resource to supply a model for performance 
>simulation to access?
>* if I develop a service that makes use of existing resources, how 
>do I need to interact with the service during development and 
>initial testing - to what extent would we like each resource to 
>supply a report mode that returns what would happen but does nothing 
>real?
>* it would appear that simulation and report mode could overlap but 
>there should probably be leeway for the owner to decide how much 
>overlap is appropriate
>* likely to have specific process models for development support
>* again raises question about what gets bundled into a single 
>service, i.e. if very diverse set, does each WSDL operation need 
>simulation and report-only modes?
>* is it appropriate to expand these questions (as I've done) to a 
>more general resource rather than only services or can we currently 
>limit resources of interest to those accessed by a SOA service?
>
>My initial questions are:
>1. to what extend is this in the RA scope?
>2. if in scope does it belong in management?  does some of it belong 
>in the Realizing view?  where else might it show up?
>
>That probably creates enough trouble for now.
>
>Ken
>
>Attachment converted: Macintosh HD:smime 766.p7s (    /    ) (006DC9D4)


-- 
Rex Brooks
President, CEO
Starbourne Communications Design
GeoAddress: 1361-A Addison
Berkeley, CA 94702
Tel: 510-898-0670