Re: New draft on embedding the RP address in IPv6 multicast address

[Bill Nickless <nickless@mcs.anl.gov>]

At 10:08 PM 10/16/2002 +0300, Pekka Savola wrote:

> Smaller segments can be used if that's really a problem (100/FD should
> probably handle some 20-30Mbit/s of multicast per port without anyone
> noticing anything but I haven't tested).
If I have to re-engineer a network, resegmenting and renumbering, just to 
support multicast (IPv4, IPv6, SSM, or ASM), then I'd better have a darn 
good business case to go to all that expense and complexity.

Taking your suggestion to an extreme, I would have to provide a layer-3 
routed port for every high-bandwidth multicast host.  Should I have to go 
to the trouble of assigning IPv4 /31s or /30s, or IPv6 /126es for every 
multicast-speaking host?  Sorry, it's just not going to happen.

IEEE 802 switches function just fine to send unicast frames only to the 
ports for which they're intended.  Unicast IP uses ARP/ND to discover the 
MAC addresses that the IEEE 802 switches depend on to make this 
happen.  What's wrong with asking IP multicast to use a similar mechanism?

===
Bill Nickless    http://www.mcs.anl.gov/people/nickless      +1 630 252 7390
PGP:0E 0F 16 80 C5 B1 69 52 E1 44 1A A5 0E 1B 74 F7     nickless@mcs.anl.gov