Pushpasis,

On Oct 7, 2015, at 3:47 PM, Pushpasis Sarkar 
<[email protected]<mailto:[email protected]>> wrote:

Hi Bruno,

From: "[email protected]<mailto:[email protected]>"
Date: Wednesday, October 7, 2015 at 5:43 PM
To: Pushpasis Sarkar, "Stefano Previdi (sprevidi)"
Cc: Hannes Gredler, "[email protected]<mailto:[email protected]>", Isis-wg, Imtiyaz 
Mohammad, "Clarence Filsfils (cfilsfil)"
Subject: RE: [Isis-wg] Handling same SID mapped to different prefixes and vice 
versa cases

BTW, draft-ietf-isis-segment-routing-extensions does not explicitly state 
whether this is allowed or not (to advertise Special Purpose MPLS labels (aka 
Reserved labels)). My personal reading is that since this is not forbidden, 
then this is allowed.
[Pushpasis] That is exactly what I (and probably Imtiyaz too) wanted the 
authors to confirm and if possible explicitly spell out in the next version (we 
expect that draft should specifically spell out that same node-sid can be 
associated with multiple prefixes originated by the same node..


is it what Bruno intended ?  Sorry, this is not what I understood. What I agree 
with is:

. multiple prefixes may use the node-sid of their originator (the indication 
may be done through Bruno’s suggestion) and through the use of prefix-attribute.

This covers the case where you want to use:
.  the same node-sid for all prefixes of the same node
. some prefixes use one node-sid and other prefixes use another node-sid
. prefixes use the node-sid while also carry a local sid.

one solution for many use cases and without opening the door to 
misconfigurations.

s.



But not with same or different prefixes originated from different nodes). 
Otherwise implementations may interpret this in other way and expect all 
prefixes (even originated  by the same node) to have separate index associated 
and if they receive two or more prefixes form same node with same index they 
might throw and error and not install the label push for the rest of the 
prefixes (i.e. program it for only the first prefix learnt).

Thanks
-Pushpasis

_______________________________________________
spring mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/spring

Reply via email to