[rbridge] Updated charter
erik.nordmark at sun.com
Fri Feb 4 07:17:22 PST 2005
Joe Touch wrote:
>> So perhaps you can list the numerous problems.
> It works fine, as you noted, over pt-pt links, but not over things that
> are intended to look like shared media, e.g., subnets.
And the numerous problems in that case are ....
> Why? It walks like a router, talks like a router, and quacks like a
> router. Examining the IP layer, changing L2 headers, etc. The only thing
> it doesn't do that a router does - so far - is decrement the TTL. Others
> would be, e.g., to limit all 1's broadcast.
It might route the packet based on the L2 address instead of IP, for
instance. Hence using a different term than "router" aids in clarity IMHO.
>> What we are talking about is a hybrid, which includes L2 semantics but
>> might not preserve L2 semantics when carrying IP packets.
> Huh? You're preserving L2 semantics but only for IP, which doesn't care
> (above) about such preservation?
I think you need to re-read what I wrote above.
> But we were talking about such support in ways that avoided broadcast...
Who has been talking about hybrid devices which interconnect broadcast
and NMBA L2s?
>> At another end of the scale we have what I'd call "IP works". In this
>> case the cloud of interconnected hybrids collectively exhibit behavior
>> so that IPv*/ARP/ND/DHCP etc work as expected.
> Why can't we call that a router that doesn't decrement TTL?
Because, AFAIK, it forwards packets based on L2 addresses.
(Remember, we had this mailing list discussion about packets sent to
off-subnet destinations and the interaction with redirects etc, which
resulted in at a minimum packets destined off-subnet need to be routed
based on the L2 destination.)
> Bridges can - and do limit the spread of multicast. They do not limit
> the spread of broadcast based on L3 semantics (unless doing explicit
> proxy-arp, but that's not what we're talking about).
I'm confused. We'd talked about ARP/ND flooding a while back, and
concluded there were security and robustness issues around that, which
But I thought the current topic was about interconnecting different
types of L2s (with different address formats), which is orthogonal to
any limitations on the flooding of broadcasts.
More information about the rbridge