Re: [TE-wg] TE use in today's networks

[Jennifer Rexford <jrex@research.att.com>]

Venkata,

> But I hardly saw in OSPF/ISIS WGs, any drafts on dynamic recomputation
> of routes based on dynamic link weights. I doubt any vendor working on
> this to practically implement (not even in foreseeable future)

I'm not sure I understand.  The idea is to use OSPF/IS-IS as it exists
already, with "static" link weights (set manually by a network
operator or automagically by a network management system).  That is,
the routers do *not* need to talk amongst themselves to select the
weights.  Rather, the weights are selected externally based on global
observations of the traffic and the topology.  This would not mandate
any changes to the existing IGPs.  That is the main positive feature
of this approach -- it doesn't require any protocol changes or any new
working group activity.  (This is not to say that I necessarily
disagree with the other approaches.  Rather, I am saying that the
other approaches need to be compared with a "default" approach of
working within the existing technology base.)

-- Jen