[rbridge] Consensus around MGROUP set of PDUs

[Ali Sajassi (sajassi)]

Hi Linda,

MGROUP PDU is control-plane related and not data-plane and thus no
correlation to the need for additional L2 mcast group addresses.

Cheers,
Ali 

> -----Original Message-----
> From: rbridge-bounces at postel.org [mailto:rbridge-bounces at postel.org]
On
> Behalf Of Linda Dunbar
> Sent: Friday, July 02, 2010 3:16 PM
> To: 'Donald Eastlake'
> Cc: 'Developing a hybrid router/bridge.'
> Subject: Re: [rbridge] Consensus around MGROUP set of PDUs
> 
> Donald,
> 
> I mean why doesn't TRILL allow many multicast addresses, like
> 01-80-C2-00-00-4x, where x can be 0~F, with each address represents
one
> multicast domain?
> 
> Or even 01-80-C2-00-00-xx, where you can have 256 multicast addresses,
> representing 256 multicast groups. There are plenty of multicast
> addresses
> to use. Using different multicast addresses to represent different
> multicast
> groups have worked very well in Ethernet networks.
> 
> Linda Dunbar
> 
> 
> -----Original Message-----
> From: Donald Eastlake [mailto:d3e3e3 at gmail.com]
> Sent: Friday, July 02, 2010 4:58 PM
> To: Linda Dunbar
> Cc: Erik Nordmark; Developing a hybrid router/bridge.
> Subject: Re: [rbridge] Consensus around MGROUP set of PDUs
> 
> Hi Linda,
> 
> On Fri, Jul 2, 2010 at 4:29 PM, Linda Dunbar <ldunbar at huawei.com>
> wrote:
> > Does TRILL have different multicast addresses for different
multicast
> > groups, like IEEE802.1Q does? Why need different PDU for different
> multicast
> > groups?
> >
> > Regards, Linda Dunbar
> 
> I'm not sure exactly what you mean. In TRILL, multi-destination data
> frames are sent between RBridges encapsulated to the All-RBridges
> multicast address (01-80-C2-00-00-40), while data frames with a known
> unicast address destination are sent between RBridges encapsulated to
> the unicast address of the next hop RBridge.
> 
> In order to be able to optimize the distribution of multicast native
> frames, RBridges need to know what other RBridges have attached
> listeners for particular multicast addresses, so they can prune the
> distribution tree. This information is distributed as part of the link
> state (along with multicast router attachment information).
> 
> Ideas with the MGROUP PDU set were to provide one more set of 256
> fragments to accommodate more link state information and to segregate
> the probably more rapidly changing multicast listener information
> (more rapidly changing than topology); however, there are other
> techniques if more fragments are really needed and more rapidly
> changing information should be in different fragments anyway so you
> don't need a different PDU.
> 
> Thanks,
> Donald
> 
> _______________________________________________
> rbridge mailing list
> rbridge at postel.org
> http://mailman.postel.org/mailman/listinfo/rbridge