|
One outright comment I have first if why not send a unicast
RA rather than sending an RA in an IPv6 packet with destination as L3 multicast
but the L2 is a unicast destination. Also, if this proposal has to
go through, the test performed are insufficient. Test MLDv2
multicast with this proposal and publish results where CAM in network hardware
is going to sniff 3333.xxxx.xxxx for MLDv2 control. Other hardware
accelerated router platforms should also be tested for such a proposal to see
what works for send and receive of such doctored packets. Lastly, it would be good to give very copious details on the
architecture when such statements are made in the document: [It is up to the system architecture as when to transmit an
IPv6 multicast message as an link-layer unicast message.] Which IPv6 deployment architecture has been accepted and
published that caters to such a proposal for changing the L2 of a L3 multicast
destined packet. Hemant |