​Hi Rob,​

> ​
> It introduces a requirement, where failure detection is required, to run N
> different OAM probes to be able to monitor reachability (one for each
> Node-SID assigned to the remote device - even in cases where we have 1
> available path).
>

​When you get today /24 IP prefix ​do you monitor reachability to all 255
host addresses within this prefix or to the next hop which advertised it ?

Likewise when you get a block of SID monitoring OEM to the node which
advertised it seems sufficient to me. I assume there is still IGP there and
router_id or s-bfd.

Comparing two options it seems that mpls hardware is to much faster support
match on subset of MPLS label bits vs ability to look over long chain of
labels for ELI/EL.

I think the question for the MPLS WG here is whether there is any appetite
> to introduce a new means by which entropy in MPLS networks is introduced.
> We already have Entropy Label which has been published and implemented. The
> question to me seems to be why we need anything else. Fewer solutions seems
> optimal :-)
>

​Of course fewer is better provided that:

a) fewer works
b) is architecturally sound

Moreover SID/label block vs entropy labels is not apple to apple
comparison.

Former is just control plane solution, does not insert anything extra into
each packet while latter is based on data plane modifications and new
hardware support to either handle Eiffel Tower of label stack or tricks
with search along the stack for ELI/EL tupel(s).

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

Reply via email to