Re: [soa-rm-ra] An epiphany

[Ken Laskey <klaskey@mitre.org>]

Dang, Frank, I go away for a couple days and you go all wacko on me :-)

OK, my take.  Of course it has nothing to do with messages.   
Information has to get from Participant A to Participant B and might  
also depend on information provided by Participants C, D, and E.   
Talking about message exchange buys us three things:

1) Everybody exchanges messages now and it gives the RA some concrete  
grounding in a space that is familiar to most people.

2) I expect there is very little if anything messages add to the RA  
than saying you need abstract (and eventually concrete) components for  
message passing and processing if you're exchanging messages, but it  
gives us the incentive to really consider if messages are more than a  
specific use of the SOA machinery that other things point out.

3) I forget right now, but there was a third.

As you note, when you access and want to use information you get from  
someone else, you have to understand it.  UDDI is based on the  
assumption that this is not an issue.  Obviously, those folks never  
went to an online vendor and tried to make use of the product  
comparison tables.

Description has nothing to do with messages, so getting rid of  
messages doesn't get rid of description.  Description does, however,  
critically need semantic engagement.  That is one of the drivers for  
the section of description Danny always wants to kill because it  
explicitly points out that whenever you use a term, you need to  
disambiguate the vocabulary being used.  This sets the stage for  
mediation which is needed all over the place if independent entities  
are independently creating and providing resources.

Your system of systems of systems looks nothing more than compositions  
at various levels and for various types of resources.  I need more  
than semantic engagement for the service input values.  I also need it  
for policies, for technical assumptions, for ... .  That is where the  
execution context comes in because it is the collection of the  
semantic engagements at various levels over the various aspects of SOA.

Now I don't see where this cuts the RA in half.  If it keeps the  
length manageable, that's a plus, but I don't yet see it.

Ken

On Jan 16, 2009, at 12:24 AM, Francis McCabe wrote:

> Hi Danny,
>  I realize that I have some explaining to do...
>
>  I don't see Counts-as as being connected to semantic engagement; at  
> least not at the moment.
>
>  Like I said, this is really about the dichotomy/relationship  
> between private internal action and publicly shared/understood  
> interaction. In my view, the messaging system is, at this level of  
> abstraction, simply a medium in which you can have such a public  
> interaction. It is necessary to have such a medium because we are  
> fundamentally talking about peer-to-peer models crossing ownership  
> boundaries.
>
>  On top of this, is the strong intuition that one can have multiple  
> layers of ecosystem -- systems of systems of systems etc.
>
>  One of the diagrams that we have drawn internally is like a spider  
> plant:
> <Spider Plant.pdf><ATT00001.htm>

-----------------------------------------------------------------------------
Ken Laskey
MITRE Corporation, M/S H305      phone: 703-983-7934
7515 Colshire Drive                         fax:       703-983-1379
McLean VA 22102-7508