<Snip> > It is also not true that uSID inflates the IGP and/or FIB tables > more than other approaches. A uSID is advertised just like > other SRv6 SIDs, although the prefix length will typically > be much shorter. The fact that no extra label mapping > table is required contributes to improved control and > data plane efficiency and provides excellent forwarding > ASIC efficiency, especially for low-FIB and legacy systems.
I don't agree with this - and here is why: To use uSID - I either have to use a locator ID in additional to my normal loopback - or - in the alternative - renumber things such that my normal loopback numbering scheme would be such that shifting on the address would not break things on the network. The latter option in a large network is entirely and totally impractical in my view - so - that means - I now need to add a second address to every router in order to use this. This doubles the number of loopbacks in the IGP for every node that I wish to use a uSID on, since I now have the original loopback - and the address that I wish to use for steering. In a CRH context - because of the de-coupling of the SID and the address - there is absolutely no problem with applying the SID's to the original loopbacks - and no increase in the IGP. Furthermore, the additional locator addresses I am now applying - because they are in effect just another v6 address - will propagate through the IGP to every single device - irrespective of if that device needs to know about it or not - this means that on the legacy hardware that has low fib counts - I will see an impact. I do not need locator SID's in addition to my normal loopbacks with CRH - so - in inflates the IGP in my eyes. In a large network - where a lot of steering is happening - and you have a large number of devices through which you are going to need to steer - this can actually have quite a serious impact. Thanks Andrew _______________________________________________ spring mailing list spring@ietf.org https://www.ietf.org/mailman/listinfo/spring
