[rbridge] Orphaned endnodes with partitioned VLANs on a cloud

[Eric Gray]

Donald,

	See some amplifcations below...

Thanks!

--
Eric Gray
Principal Engineer
Ericsson  

> -----Original Message-----
> From: rbridge-bounces at postel.org 
> [mailto:rbridge-bounces at postel.org] On Behalf Of Eastlake III 
> Donald-LDE008
> Sent: Monday, December 10, 2007 10:58 PM
> To: Developing a hybrid router/bridge.
> Subject: Re: [rbridge] Orphaned endnodes with partitioned 
> VLANs on a cloud
> 
> The incremental deployment requirement means that the links between
> Rbridges could be arbitrarily complex bridged LANs including hubs and
> repeaters. A physical link or hub or repeater or bridge coming up or
> failing within such a bridge LAN could variously increase or decrease 
> or partition or unite the sets of Rbridges interconnected by such a
> bridged LAN, in some cases due to spanning tree reconfiguration.

It is at least likely that some of these may occur as a result of IS-IS
RBridge peer discovery as well - hence the need to ensure that we define

things correctly (and - hence - the reason why I wanted to define all
such things within a single LAN/VLAN context) so that RBridges - and 
existing bridges - work correctly as they would in a comparably deployed
network of either all bridges or all RBridges.

Note the the "comparably deployed" is also very significant to the goal
of incremental deployment - since that allows the simplest transition
models (1 by 1 replacement).

> 
> (I must also say that I think the word "media" has mostly to do with
> link physics in my mind. Whether a link can be a bridged LAN or not
is,
> in my opinion, a wholly different type of question from whether it is
> any particular type of "media" such as wired or optical 802.3, 802.11,
> 802.16, etc., etc. I don't think anything in the current TRILL spec
> depends on what type of physical media your are using, it only
requires
> that it be "802.1 conformant". How to handle bridges and whether they
> are allowed is pretty independent of which or what combination of the
> above "media" types you have.)

I think I agree, though I feel that I may be missing the context.  What
part of the thread below prompts this parenthetical observation?

> 
> Thanks,
> Donald
> 
> -----Original Message-----
> From: rbridge-bounces at postel.org 
> [mailto:rbridge-bounces at postel.org] On
> Behalf Of Russ White
> Sent: Friday, December 07, 2007 12:44 PM
> To: Radia Perlman
> Cc: Developing a hybrid router/bridge.
> Subject: Re: [rbridge] Orphaned endnodes with partitioned VLANs on a
> cloud
> 
> > Not trying to "fix" the failure mode is exactly what the 
> current thing
> > is doing: it favors simplicity and scaling, and not
> > trying to heal partitions within a cloud, but instead 
> making sure that
> > if there is a partition, it won't cause loops.
> 
> Loops on the edge ports are a problem--I'm cogitating on a set of
> solutions for that which don't require the complexity involved in the
> proposed solution. IMHO, the specification should be flexible 
> here, and
> allow tradeoffs.
> 
> I'm not getting where a loop can form due to a link failure in the
> underlying transport between rbridges. Could someone provide 
> an example?
> 
> :-)
> 
> Russ
> 
> - --
> riw at cisco.com CCIE <>< Grace Alone
> 
> _______________________________________________
> rbridge mailing list
> rbridge at postel.org
> http://mailman.postel.org/mailman/listinfo/rbridge
>