Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

RE: [EFM] Wavelength allocation




The big issue here is that CWDM is still an expensive thing to imeplement
compared to p2p or EPON. Sure one can buy an off-the-shelf eight node 1GE
CWDM sub-system, but the business case doesn't scale to the fat part of the
market. CWDM really needs tunables too (so that spares are less of an
issue), and that is a real challenge economically at the moment.

It will take a couple of years, and some market pull, to bring more vendors
into the CWDM component and sub-system business. One way of creating 'pull'
and at the same time create a multi-vendor supply chain for CWDM would be to
write an EFM standard around CWDM wavelength allocation. I for one would
support such an effort.

Bob

-----Original Message-----
From: owner-stds-802-3-efm@majordomo.ieee.org
[mailto:owner-stds-802-3-efm@majordomo.ieee.org]On Behalf Of Carlos
Ribeiro
Sent: 17 July 2001 12:20
To: carlosal@ctbctelecom.net.br
Cc: stds-802-3-efm@ieee.org
Subject: [EFM] Wavelength allocation



I've tried to send this message a few times over the weekend, but I didn't
get through. I'm trying again, because this is a very important topic,
looking from the carrier's perspective.

Most EPON proposals are P2MP. You have to use a sophisticated arbitration
mechanism to control the use of the upstream channel. Although complex and
initially expensive, DOCSIS did this before, and with high volume the cost
of CPEs is driven down. However, the user is still concerned with the
security (ok, we know it's almost impossible to tap the neighboor upstream
signal, but go figure customer's mind :-). Also, the amount of bandwidth is
limited by design. Any changes must be implemented by all elements in the
give cable/fiber segment.

In the P2P EPON proposal, each customer gets their own wavelength. You have
PON at the physical level, but a point-to-point connection at the upper
layers. I believe this is the best of both worlds, because you have high
speed, and you can provide dfferentiated services for some customers in the
segment. You're not stuck with the technology of the traditional PON
central node.

As the technology evolves, it's possible that we'll hit the technical
limitations of the first generation EPON systems - the ones that are going
to be deployed in the next months. In the cable industry, you use smaler
segments to improve efficiency. Using P2MP EPON, you can do the same, but
you can also start to deploy new wavelengths for the new services. In fact,
it is possible to envision a scenario where you end up allocating a
wavelength for every individual customer, turning your P2MP network into a
P2P network.

Other applications can (and should) user their own wavelengths. The best
example that comes to mind is broadcast video - it's much more efficient
and less expensive to deliver it using a separate lamda, than trying to
carry it over packets.

As it is, I would like the EFM standard to include a wavelength allocation
policy. I haven't seen this discussed in depth; some presentations touch on
this topic slightly. There is someone working specifically on this?


Carlos Ribeiro
CTBC Telecom