[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

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



> i agree with Bill's earlier comment: lack of IGMPv3 snooping in
> current switches is a major obstacle to IPv4 SSM deployment, and
> lack of MLD snooping will become an impediment to IPv6 SSM
> deployment.

I've said it before and I'll say it again: it strikes me as
architecturally wrong to have traditionally cost-sensitive layer-2
switching devices having to implement interpretation of layer-3 protocol
information in frames not explicitly directed at itself.  It's IGMPv3
this day, and MLD for IPv6 the next day, and whatever we might dream up
next after that.  Multicast protocol development appears to still be in
flux.  This makes for complexity which leads to deployment problems
because old non-field-upgradeable devices need to be replaced and one
needs to wait for switch vendors to become aware of and then to
delivering software upgrades (even if the devices are field-
upgradeable).

It would seem to me to be an architecturally better and longer-term more
"stable" way to put the layer-3 smarts in the layer-3 devices, and to
devise a protocol where the layer-3 devices could direct the layer-2
devices as appropriate, a'la Cisco's CGMP.

Regards,

- Håvard