All things related to multicast

[John Kristoff <>]

On Sat, 10 Aug 2002 00:15:38 +0200 (CEST)
Havard Eidnes 
<>
 wrote:

> Indeed.  Having the L2 switch implementors need to track and implement
> the required intelligence to support the various variations we come up
> with at L3 over the lifetime of the L2 devices has always struck me as
> architecturally awkward and unlikely to produce the most economical L2
[...]
> Correct me if I'm wrong, but I thought that this is part of what the
> Cisco-specific CGMP protocl does?

I'd echo Havard's comments.  What are the drawbacks with a standards
based cgmp mechanism as opposed to snooping?

The one major requirement for a network protocol such as cgmp that
snooping doesn't require is a authentication mechanism between the
communicating devices.  Shared keys, encrypted in transmission should
suffice.

John