Cisco developers have, and continue to make, major contributions into
what I hesitate to call MPLS. Some context may help here.
First, one has to understand that protocol families like MPLS do not
cleanly fit into the traditional model, and you can't force-fit them.
When I say "traditional" model, however, I'm restricting that to the
seven-story apartment house of ISO 7498. MPLS fits better with the
revisions in the ISO document "Internal Organization of the Network
Layer".
Even beyond that, however, the IETF struggled with how to handle
these and related protocols, and eventually set up the "sub-IP"
directorate -- intelligent transmission systems below IP but far more
complex than traditional data links. This isn't restricted to MPLS,
but also covers IP over optical, IP over cable, generic switch
management protocol, and daughter-of-MPLS, Generalized MPLS (GMPLS).
Cisco educational materials have long overemphasized the forwarding
part of MPLS and sort of assumed "here a miracle happens" regarding
path setup. I remember trying to teach a beta class on MPLS on the
ex-Stratacom 8850, turning off the projector, turning to the class of
Cisco SE's, and going to the whiteboard to spend 45 minutes
introducing how MPLS actually worked.
In particular, the roles of MPLS signaling protocols such as basic
LDP, RSVP-TE, and extended LDP were skimmed over, and the dependence
of these protocols on conventional IP routing was minimized. Little
attention also was given to the extremely rich traffic management and
high availability features of MPLS, which I consider the main
motivation for using it -- not forwarding performance improvements,
which, at best, are minimal.
Truly understanding the direction of these technologies works much
better when you understand the generalization of GMPLS and see how it
gives a common way of dealing with traditional technologies. Up to
now, MPLS was packet/frame oriented. The GMPS extensions, however,
allow you to use a largely common control framework for:
packets/frames
wavelengths (lambdas) in pure optical networking
timeslots in TDM networking
port identifiers when working with DACS and the like.
I can't necessarily recommend any pure MPLS books, because I go
directly to the IETF documents when I need to check something -- and
am on the developer mailing lists. There is a significant amount
about ISP applications of MPLS, however, in my book, _Building
Service Provider Networks_ (Wiley, 2002, ISBN 0-471-09922-8), for
which our own Annlee Hines was my peer reviewer, and Scott Bradner
and Lyman Chapin were advisors.
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=66698&t=66609
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
