[rbridge] Orphaned endnodes with partitioned VLANs on a cloud

[Anoop Ghanwani]

> -----Original Message-----
> From: Russ White [mailto:riw at cisco.com] 
> Sent: Thursday, December 06, 2007 10:03 PM
> To: Anoop Ghanwani
> Cc: Radia Perlman; Developing a hybrid router/bridge.
> Subject: Re: [rbridge] Orphaned endnodes with partitioned 
> VLANs on a cloud
> 
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> 
> > Rbridge core
> > |         |
> > RB1      RB2
> > |         |
> > +---------+
> > 
> > This is the case...RB1 and RB2 are connected to the core on the top 
> > and at the bottom they are supposedly edge ports.  However, these 
> > accidentally get connected.  What we have is loop and multicast 
> > traffic which gets sent into the core by both RB1 and RB2.
> 
> Two things:
> 
> 1. There are other ways to resolve this. uRPF would probably 
> would, in fact. Or, perhaps, sending out hellos marked so 
> they don't form an adjacency in IS-IS. I should point out 
> that modern bridges don't seem to have a problem with the 
> bpdu transmissions in this situation, so I don't see why they 
> would have a problem with IS-IS hellos.

Because they don't have to send 4000 BDPUs if they have
4000 VLANs configured on that port.

> 2. We seem to tolerate this misconfiguration problem in 
> modern day switches all the time, and everyone just seems to 
> live with it.
> 
> I would point out that the link here is not intended to 
> transit traffic.
> I'm waiting for an example where you'd have 4000 VLANs 
> connected between two rbridges which are intended to transit traffic.

Just the same example above.  RB1 & RB2 are edges to the
RBridge core but they are each connected to bridged LANs
configured with 4K VLANs.

Anoop