Hi steve,

> >In the future I can see two cases:
> > - the flow label is set by the source (first camp)
> > - the flow label is set by an edge router (with the
> DiffServ definition
> >   of what is an edge router) because the source box (or its user) is
> >   too dumb to deal with QoS/..., according to the edge
> router manager.
>
> If the Flow Label is set by something other than the source node, how
> do you guarantee its uniqueness, over the topological scope in which
> it must be unique?  Do you need to invent a protocol between
> the routers
> on a LAN to divvy up the Flow label space?

It could be set as the MPLS protocol sets its traffic flows and distributed
to how MPLS distributes its flow info in its domain (RSVP-TE, LDP). This was
what I have had in mind.

And you would utilize the flow label inside the ip header and get rid of the
intemediate L2.5 protocol (MPLS).

The intention was of course to map the traffic onto a LSP in the edge router
inside your domain..

This could be done i a label stacked approach to which we will need the
dest. option Jim was talking about. The host sets the flow label for the
session as propose by Itojun and Jochen's approach. Into the backbone/core
of your network we might want to do traffic engineering and to use the flow
label for that purpose. And if we have one bit telling that the flow label
should be distributed to the other end, we need a dest option for the
original flow (which had a semantic in the access) that needs to be swapped
back at the egress of your domain. Then we have the "new" flow label in the
core that we do TE on.

> What happens when the LAN
> partitions, then routers in the different partitions assign duplicate
> Flow Labels, and then the partition heals?

It is solved exatcly as it is solved in mpls in the core and in the access
we would need to integrate admission control in the access router.

So the host signals for a flowlabel to the access router that do admission
control and also verifies the uniqueness of the flow-label in the access...


-- thomas




--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to