RE: Application Protocol Definition (was RE: Visibility (was Re: Introducing the Service Oriented Architectural style ...

David,
 
If we classify protocols in this way, what should we call ebXML and
RosettaNet as a whole? In other words, what category name(s) should be
given for ebXML and RosettaNet types of protocols?  We may call ebXML
messaging and RosettaNet messaging as Messaging protocols, ebXML
business process and document types and RosettaNet PIPs as Business
protocols, ebXML Registry as Registry protocol,  and so on.  
 
Protocols, as objects,  have all the object relationship.   So I don't
know  if we can just try to classify them neatly.  I understand
"Application Protocol" is vague, but IMHO it might be easier to define
what scope is for an Application or what an "Application Protocol"
really means here.
 
This is just my thought.
 
Jenny
 
 
 
 

-----Original Message-----
From: Burdett, David [mailto:david.burdett@commerceone.com]
Sent: Thursday, February 27, 2003 6:34 AM
To: 'Assaf Arkin'; Mark Baker; Burdett, David
Cc: www-ws-arch@w3.org
Subject: Application Protocol Definition (was RE: Visibility (was Re:
Intr oducing the Service Oriented Architec tural style ...



Arkin 

You said ... 

>>> This of course doesn't help much, because we've just introduced
three different protocols, one described by HTTP, one described by WSDL
and one described by WSCI (just as an example). Again as with
synch/asynch, this is all a matter of applying a definition in the
proper context.<<<

It might not help, but it is real life. You do have multiple protocols
operating in combination at different "layers", for example (not sure
this list is complete or the names are right)

1. Network infrastructure protocols, e.g. TCP/IP 
2. Communication protocols, e.g. HTTP 
3. Messaging protocols, e.g. ebXML Messaging or WS-RM 
4. Business protocols, e.g. Order placement protocol that defines
sequence of exchange of business documents 

You can also have other more slightly more specialized but still general
"protocols", e.g. 
5. Two phase commit, e.g. Business Transaction Protocol 
6. Discover protocol, e.g To discover a WSDL or Schema definition (REST
can work well here) 
7. Negotiation protocol, e.g. To negotiate which combination of
protocols 1 through 4 to use in a specific instance 

... we could probably go on ... 

The point is just calling some or all of these an "Application Protocol"
is, IMHO, insufficiently precise. 

Do you think we should try and define these types of protocol and what
they mean in more detail? 

David 

-----Original Message----- 
From: Assaf Arkin [ mailto:arkin@intalio.com] 
Sent: Wednesday, February 26, 2003 12:13 PM 
To: Mark Baker; Burdett, David 
Cc: www-ws-arch@w3.org 
Subject: RE: Visibility (was Re: Introducing the Service Oriented 
Architec tural style, and it's constraints and properties. 


> On Wed, Feb 26, 2003 at 03:18:48AM -0800, Burdett, David wrote: 
> > <DB>We need to define what we mean by an "application" if you mean
it is 
> > anything above the transport layer, then you are correct but 
> really I think 
> > the layers are typically: Operating System, App Server, "Web
Services 
> > Middleware", Application. 

How about: 

Application - A program designed to assist in the performance of a
specific 
task, such as word processing, accounting, or inventory management 

Now the only question is 'what application are we talking about?' 

Are we talking about the HTTP or FTP server? In this case HTTP and FTP
are 
the application protocols. 

Are we talking about accounting? In this case the accounting protocol is
the 
application protocol. 

Is it possible to have an application on top of an application on top of
an 
application? How about my accounting application running inside a WS 
container (in itself an application) implemented inside an HTTP server
(in 
itself an application). Is that possible? 

This of course doesn't help much, because we've just introduced three 
different protocols, one described by HTTP, one described by WSDL and
one 
described by WSCI (just as an example). Again as with synch/asynch, this
is 
all a matter of applying a definition in the proper context. 

arkin 

> 
> The *critical* thing that one has to accept in order to understand
REST, 
> is that application protocol methods are the same as operations in an 
> API, i.e. at the same layer of the stack as "getStockQuote" or 
> "purchaseBook".  If you just take this as a given for a moment, you'll

> see that all the arguments I've ever made on this subject become a
big, 
> complex, yet entirely self-consistent description of much of Web 
> architecture, and indeed several other Internet scale architectures.
If 
> you don't accept it, then I probably come off as a loon, which I 
> completely understand because I thought the same thing of some guys
who 
> saying that to me back in 97/98 (Dan Connolly and Roy Fielding, FWIW).

> 
> So, a *rhetorical* question for those of you who don't believe that 
> GET is at the same layer as getStockQuote; what would you call a 
> protocol that does have a "getStockQuote" method?  Note; "application 
> protocol" is already taken. 8-) 
> 
> MB 
> -- 
> Mark Baker.   Ottawa, Ontario, CANADA.        http://www.markbaker.ca 
> Web architecture consulting, technical reports, evaluation & analysis