Re: [v6ops] draft-ietf-dhc-pd-exclude

["Hemant Singh (shemant)" <[e-mail]>]

 

From: Lorenzo Colitti [mailto:[e-mail]] 
Sent: Wednesday, December 07, 2011 5:35 PM
To: Hemant Singh (shemant)
Cc: jouni korhonen; Thomas Narten; Ray Hunter; [e-mail]
Subject: Re: [v6ops] draft-ietf-dhc-pd-exclude

 

>Wait, what?

 

>You have a set of PD clients on the same link and want to exclude the
same /64 from each of the PDs you hand out? That doesn't make sense.
Since each of the PD clients will >get its own prefix (say, a /60) and
they won't overlap (since they're different customers), only at most one
of the PDs will include the /64. So there's no need to exclude >anything
for the others.

 

The DR has one network interface serving, say, 20K clients who are RR's.
DR is a CMTS (access concentrator) and the RR's are IPv6 CE routers
behind bridged cable modems.  Like the use case mentioned in the
pd-exclude document, the DR is the next-hop to the RR.  The DR is using
a /64 on the network interface and the DR issues multicast RA from this
network interface.  The /64 matches as the exclude prefix for one RR but
the DR happens to use the /64 on its interface multicasting the RA to
all of the 20K clients.  Thus the /64 is off-link to all clients and
"excluded".  Also, none of the other 19,999 RR's need to support the
pd-exclude option.   Thus soon as the SP domain gets one RR's pd-exclude
to say, a /64, then the SP provisioning does not support pd-exclude RR's
on the same network interface on the SP DR.  

 

Alternatively, does the DR uses unicast RA to each of the 20K clients.  

 

Do folks think such details needs to be added to the pd-exclude document
as an Appendix so that folks have some SP provisioning background.

 

Hemant