MPLS and NETLMM (was RE: a couple of high-view questions)

James Kempf wrote:
> MPLS is not on our charter. Let's not speculate on
this right now.
> When it comes time for rechartering, we can discuss how
MPLS could be
> used.

I agree completely.

> It should be sufficient that the base protocol will
support other
> than per MN tunnels.

Ok, but will it work ok when there are no tunnels? (no MPLS
paths, no
GRE tunnels, no IP-in-IP tunnels).

The protocol does not assume MPLS - ok, but it does seem to
assume that
there are tunnels of some sort.  The logical topology is:

                lma
               / | \ tunnels
              /  |  \
             AR AR  AR

When links between ARs were proposed, or multiple LMAs, it
was agreed
that it would be a load-balancing optimization for later, so
below
topology is to be supported later:

               lma
             /  | \
            /   |  \
          AR---AR--AR

How about this physical bus topology:

               lma
                |
            ---------
            |   |   |
           AR  AR  AR

In this latter case, will communication from mobile under
one AR to
mobile under another AR always go through LMA? Is this also
an
optimization one can look at later? I doubt it can be done
easily if
the base protocol always uses tunnels to LMA.

And I agree with the main point that the base protocol
should work ok 
with MPLS tunnels and IP-in-IP tunnels and any other
underlying 
tunneling mesh.

Alex
_______________________________________________
netlmm mailing list
netlmmngnet.it
https://vesuvio.ipv6.cselt.it/mailman/listinfo/netlmm