RE: Role of intermediary

With respect to intermediaries
At 03:18 PM 1/26/01 +0000, Williams, Stuart wrote:
> > -----Original Message-----
> > From: marwan sabbouh [mailto:ms@mitre.org]
> > Sent: 25 January 2001 15:55
> >
> > Hi Stuart;
> >
> > I see where you are heading with this.  Am I correct to say that you
> > view  the intermediary as the processing element that apply some sort of
> > transformation to that incoming message?  Does the requirement
> > specification  support this notion?  Please don't get me wrong, I am not
> > advocating one way or another.  I'm just trying to get the definition
> > down.
> >
> > Marwan
>
>Hi Marwan,
>
>I think many of us are just trying to get the definition down too... I think
>that I too am more in the asking questions mode than asserting answers. So
>this is merely one persons viewpoint.
>
> > Am I correct to say that you view  the intermediary as the processing
>element that apply some sort of
> > transformation to that incoming message?
>
>Mostly yes... although I think I could also have multiple distinct
>viewpoints. I tend to think of an XP Module as having both syntactic
>elements (the XP Blocks) with their semantics and a set of rules that govern
>any processing of those blocks. One could view a message passing through
>entities that applied the rules of a given XP module to the XP blocks
>associated with that XP module. Those entities could exist anywhere between
>the sending and receiving XP processors. So an intermediary would be a
>container (an execution environment) for those entities.

I strongly agree with Stuart's direction here. Marc Shapiro wrote a seminal 
paper regarding this "intermediary chaining" approach to distributed 
computing a number of years ago. You can find it here: 
<http://www-sor.inria.fr/publi/FLEX_rr2007.html>.

One of the draft scenarios, DS6, talks specifically about encryption. Note 
that in an intermediary chaining model, encryption is essentially just 
another quality of service provided by the binding -- the chain of 
intermediaries -- logically connecting the sender to the receiver. 
Encryption, compression, transportation, marshaling, and even receiver-side 
dispatching all act as intermediaries in the chain. This model is extremely 
flexible and upgradable, and yet when implemented correctly it works very 
efficiently in practice.

--steve

======================================================
Steve Vinoski                      vinoski at iona.com
Chief Architect & Vice President Platform Technologies
IONA Technologies, Inc.          Waltham, MA USA 02451
200 West St.        http://www.iona.com/hyplan/vinoski/

Received on Friday, 26 January 2001 11:01:11 UTC