The MPLS WG Archive

[Daniel Awduche <awduche@UU.NET>]

Dimitry,

Thanks or the note. See additional comments inline...

On Wed, Apr 19, 2000 at 10:52:03AM -0400, Dimitry Haskin wrote:
> Dan,
> 
> Let me suggest the following text:
> 
> "To be definite, in the original RSVP protocol [3], a session 
> was defined as a data flow with a particular destination and 
> transport layer protocol.  In the RSVP-TE specification, however,
> session is defined as the set of label switched packets with
> a particular destination, tunnel id, and extended tunnel id."

This is a good try, but it conveys an incorrect impression
that the parameters of the RSVP-TE SESSION object (tunnel end 
point address, tunnel id, and extended tunnel id) are in fact 
attributes of the label switched packets. That is, it
confuses the semantics of the RSVP-TE signaling messages that
instantiate a session with the attributes of the packets 
that comprise the session.

Nonetheless, you did have a valid point in your original
email, in that you discerned that the wording of the 
applicability draft did not express the fact that a session 
may result in more than one LSP. With this mind, 
here's a corrected version of the second sentence which 
accommodates your input.

"In the RSVP-TE specification, however, a session is defined 
as the set of label switched packets that are instantiated
by RSVP-TE messages containing particular values of the
SESSION object parameters (Tunnel-End-Point-Address, 
Tunnel-Id, and Extended-Tunnel-Id)."

Let's converge on these rhetorical aspects and move on
to other issues... 

George/Lou feel free to step in if a more concise description 
is needed.

Cheers,
/Dan