The MPLS WG Archive

[Grenville Armitage <gja@research.bell-labs.com>]

Sanjeev Rampal wrote:

	[..]
> This enables merging operations
> etc.
> 
> In contrast optical switches work at the granularity of
> optical channels/ lambdas/ STSs which are not
> "mergeable"

And my historical note is that "merge" wasn't even
a mandatory requirement in MPLS when we first started
bringing it into the ATM world. So this is not a point
of fundamental difference, and does not invalidate the
(now highly overstretched) analogy.

Stepping  back a little....

What also seems to be missed here is that the analogy
between IP/ATM history, and IP/optical today isn't the
technical details. It is that some class of operators
emerged in the IP/atm world who believed their ATM nets
served only one client (primarily) and therefore it
made sense (for them) to optimize the ATM control for the
IP 'client' (aka MPLS/ATM).

Does this support or detract from arguments for the peer
model? The absence of large operators stating on these lists
that they're rolling out IP-only switched optical nets would
seem to favor development of the overlay model for data/optics.

cheers,
gja