W3C home > Mailing lists > Public > xml-dist-app@w3.org > February 2001

INT: Re: Intermediary Discussion

From: Scott Isaacson <SISAACSON@novell.com>
Date: Thu, 01 Feb 2001 10:05:30 -0700
Message-Id: <sa7934f1.094@prv-mail20.provo.novell.com>
To: <marting@develop.com>, <xml-dist-app@w3.org>
I think back to the very successful model of text processing and shell scripts in the early days of Unix.  All communication was abstracted in the form of stdin, stdout, and stderr.  The internal logic of a certain application didn't care if it were on orginating entity, or an intermediary, or a final processing entity, it had a job to do and it did it.  It didn't need to know where it was in the pipeline, it just needed to rely on a very standard way to get the data in, procesess it, can get it out.

If it was invoked with a command line argument it was an originating entity.
If it was invoked in the middle of a pipe it was an intermediary.
If its output was redirected to a file it was a final destination entity.

I feel like an XP intermediary is only an intermediary in how it is invoked or where it sits in the XP message path, not in its function.  

There will be a *few* type 3 (see the orginal email message below) intermediaries:

1) Suppose there is an XP Intermediary that maps from Underlying-Protocol-A to Underlying-Protocol-B and there is another that maps from Underlying-Protocol-B to Underlying-Protocol-C.  They should be chainable to essentially build a path that XP Processor 1 bound to Underlying-Protocol-A can use to send messages to XP Processor 2 bound to Underlying-Protocol-C.
2) I can see QoS bridging type 3 intermediaries
3) I can see encryption and security management intermediaries.
4) I can see negotiation intermediaries (I need to find an X server, but I don't know where to look, I will just send to you and you find it for me).

I see *many* type 4 intermediaries

1) Application to Application mapping
2) Some service X expects input in the form of X-Input but some service user Y only speaks Y-Input so there needs to be a translation of some sort
3) Shema mapping intermediaries (common name <-> user name)
4) Store and forward type intermediaries that map between tightly coupled and loosely coupled systems ("please hold" type operater service)
5) b2b financial transaction management intermediaries
6) service aggregation intermediaries ("I can do X, Y, and Z since I front end an X and a Y and a Z and in fact I can to W since W requires both an X and a Y at the same time)
7) Intermediaries that support Accouting, Auditing, and Authorization services
etc
etc


Scott
Scott A. Isaacson
801.861.7366
sisaacson@novell.com 

>>> "Martin Gudgin" <marting@develop.com> 02/01/01 01:43AM >>>
There has been significant discussion ( and disagreement! <g> ) about the
term 'intermediary' with respect to XP. This mail is intended to stimulate
discussion of this term with the goal of having a new proposal for a
glossary definition by Feb 6. If you start any new threads on this topic,
please prefix them with 'INT:' so people can pick them out easily amongst
the other traffic on this list.

The glossary[1] currently defines the following term



XP intermediary

An application that can act as both an XP sender and an XP receiver with the
purpose of forwarding an XP message along an XP message path.




Various people have commented that intermediaries come in various forms.

1.    Intermediaries in the TCP stack. For example standard IP routers.

2.    Intermediaries in the HTTP stack. For example an HTTP caching proxy.

3.    Intermediaries in the XP 'stack'. For example ??????

4.    Intermediaries in the 'B2B stack'. For example a brokerage that acts
as a go between for purchasers and vendors.

I think it would be useful to provide definitions for all of the above as it
makes explicit which kind of intermediaries are in-scope for XP and which
are not. However, providing example intermediaries for the XP 'stack' is
difficult as at this stage we don't know for sure what will be part of the
XP core and therefore do not know which services may need to be provided by
intermediaries.

I'd like the discussion try to formulate the following;

a) proposed definitions of 1-4 above

b) Examples of intermediaries in 1-4 above ( beyond those already given )

I'd rather have some discussion before proposing defintions for 1-4 but if
people feel they can't operate that way please let me know and I'll provide
some definitions for us to kick around.

Flame away,

Gudge

[1] http://www.w3.org/TR/xp-reqs/#N3030 
Received on Thursday, 1 February 2001 12:06:12 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:58 GMT