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

RE: [EFM] Active Architectures




Francois, thanks for the comments and you bring up some interesting points.
Adding copper into the conduit could allow the power to be supplied to the
node via the ONUs as an option for some deployments. Lots of possibilities
here with individually-powered ports and mix and match port speeds per
customer. There are certainly some issues that need analysis but it has some
interesting possibilities.  On the 10:1 split ratio, not only would it be
more friendly to open access, but it also implies a cheaper switch since
less queuing memory is needed. But as you mention, non-blocking is essential
to insure inherent fairness across service providers. No policing (or
policing of the police) is necessary. It wouldn't be possible to cheat.

ONU's could be designed to power up on a default "provisioning" VLAN channel
and this would display the service provider selection interface to the
customer. The customer then selects their SP and is re-mapped to that SP's
VLAN ID. This would allow individual SP's to manage their customer's ONUs
without higher layer connectivity, yet would also allow for the possibility
of managing the customer's interior home network if the customer subscribed
to such a service. This could even be via yet another service provider on a
different VLAN. Granted this is outside the scope (maybe we should start
using "OTS" as an acronym since this phrase is mentioned so much on the
reflector:)), but it has to be addressed. EFM just collects dust in a museum
if all these OTS issues aren't solved concurrently.  

-----Original Message-----
From: Francois D. Menard [mailto:f.menard@muni-ims.qc.ca]
Sent: Tuesday, August 21, 2001 7:15 AM
To: 'Curtis Knittle'; 'Horne, David M'; stds-802-3-efm@ieee.org
Subject: RE: [EFM] Active Architectures


> I think the architecture you describe merits further consideration.  

Not only does it need further consideration, but it also should serve as
a benchmark against other pure PON approaches from a cost/benefit
perspective.  Perhaps the Half-PON approach could also be equipped with
both Cat3, Cat5 and Fibre interfaces, so we have a half-PON beast with
short reach (100TX) and long reach copper (EFM) legs as well.  This
would make the device suitable for use both in a MDU and in the OSP.   I
would even say that rather than having a 16:1 split ratio, a 10:1 would
be even more appropriate as it would make the upstream entirely
non-blocking for that node, thus making this even more friendly to open
access considerations.

-=Francois=-