[rbridge] multiple ports on same link

[JR Rivers]

Seems like you'd just add the portId... that way the parallel links can be used as part of the ECMP.  Clearly, if there is 802.3ad trunking occurring on the links, they would be reported as a single logical link.

JR


On Jan 6, 2010, at 3:49 PM, Radia Perlman wrote:

> Yes, I agree we need such a portId in the hello.
> 
> Additionally, we might say something like:
> 
> "If on port pa, you see a Hello with System ID=yours and port = pb, 
> then if pb> pa, stop
> transmitting anything (including Hellos) on pa (until Listen Time 
> expires on port pa)."
> 
> Perhaps at some point someone might want to have an implementation that
> load splits encapsulation/decapsulation (based on VLANs) among such parallel ports,
> but as long as doing that is invisible to the other RBridges on the link, I don't
> think we need to specify how to do that.
> 
> So I think either should do
> 
> a) just add the field portID in the hello, and not say anything more
> b) or, in addition to adding the field, saying to stop transmitting
> anything on all but one of such parallel ports, with something like
> the wording above.
> 
> Radia
> 
> 
> 
> 
> 
> Jeff Pickering wrote:
>> 
>> 
>> 
>> RB1 might have
>> 
>>   one set of ports, say { p1, p2, p3 } on one link, and another set of
>> 
>>   ports { p4, p5 } on a second link, and yet other ports, say p6, p7,
>> 
>>   p8, that are each on distinct links.  Let us call a set of ports on
>> 
>>   the same link as a "port group".
>> 
>> 
>> 
>> Ports dont know which port group they are in a-priori since this is 
>> determined
>> 
>> by external connectivity. Port group membership can only be determined by
>> 
>> examination of received pdus.
>> 
>> But since the hello pdus sent out all ports are currently identical, 
>> (if all
>> 
>> ports have the same mac, which is most common), there is no way for a 
>> port to
>> 
>> determine which "port group" it is in.
>> 
>> If we add a portId, like in a bpdu, to the trill hello this problem 
>> could be solved.
>> 
>> The spec currently doesnt mention such an identifier. What is the 
>> thinking here?
>> 
>> 
>> 
>> Jeff
>> 
>> 
>> 
>> ------------------------------------------------------------------------
>> 
>> _______________________________________________
>> rbridge mailing list
>> rbridge at postel.org
>> http://mailman.postel.org/mailman/listinfo/rbridge
>> 
> 
> _______________________________________________
> rbridge mailing list
> rbridge at postel.org
> http://mailman.postel.org/mailman/listinfo/rbridge