ietf-openproxy
[Top] [All Lists]

[midtax] Re: Midtax comments

2001-03-09 15:50:31

I think there is a much simpler way to distinguish the scopes of Midtax and
OPES.  Midtax is involved with packet networking intermediaries, while OPES
is involved with content networking intermediaries.  The difference lies in
the end to end nature of each.  Packet networking is dealing with the the
end to end delivery of bits through packet network elements.  OPES is
dealing with application level semantics, is the end point in the formal
IETF definition from a Midtax point of view, and will make the necessary
application relays (i.e. proxies) as necessary to complete the application
transactions.  OPES uses packet networking as its communications
infrastrucuture, in which Midtax may be participating in keeping the end to
end nature coherent.

My .02

Gary

-----Original Message-----
From: Melinda Shore [mailto:mshore(_at_)cisco(_dot_)com]
Sent: Friday, March 09, 2001 12:51 PM
To: Mark Nottingham; midtax(_at_)lists(_dot_)hactrn(_dot_)net; Michael W. Condry
Cc: Maciocco, Christian; Yang, Lily L; Condry, Michael W; Bhutani,
Manasi; Erickson, Rob; Menon Rama R; ietf-openproxy(_at_)imc(_dot_)org
Subject: Re: [midtax] Re: Midtax comments


> OPES is defining an intermediary architecture for applications level
content
> management services. Owners of the "Ends" are aware these services are
> being performed explicitly requesting or allowing them.

That may (or may not) be the case with midcom,
as well.  What we're doing, though, is trying to
craft an environment in which packets really
do flow end-to-end as far as the application is
concerned.  It may take some futzing around with
a middlebox (opening a pinhole in a firewall, for
example) before the data start flowing.  A
conferencing endpoint, for example, may take a
look at its own signaling streams, get authenticated
and authorized, and tell the firewall what address/
port tuples it needs to allow to pass.  Or it may
be the case that some sort of call control server
lies in the call signaling path anyway, and it can
handle dealing with the middlebox.  In any event,
we do not assume that packet flows terminate on
the middlebox.

Melinda


<Prev in Thread] Current Thread [Next in Thread>