[c-nsp] MPLS best practices question

[Mark Tinka]

On Wednesday 23 June 2010 10:27:24 pm Oliver Boehmer 
(oboehmer) wrote:

> Hmm, IGP/LDP sync addresses a different issue than
>  GR/NSF? I would consider either "IGP/LDP sync" or "LDP
>  session protection" (either one or both) to be best
>  practices.. I personally find LDP session protection
>  more straight-forward to address the link-up blackhole
>  issue than IGP/LDP sync..

I probably should have added more meat :-):

Our point of view is in the case where there has been LDP 
failure due to IGP failure (which has resulted from one or 
more problems, e.g., link failure, node failure, software 
crash, hardware failure, e.t.c.). For us, this is a more 
common scenario than a typical case where LDP-IGP 
Synchronization would be useful, e.g.:

	- Failure of LDP Hello exchanges yet the IGP is alive and
	  well.

	- Formation of a new IGP link that has converged re: the
	  IGP, but is still building FEC's re: LDP.

After studying our network performance over a 12-month 
period across the variety of platforms we have, we concluded 
that the above 2 scenarios were sufficiently rare that 
IGP/LDP Synchronization didn't make sense for us (still 
doesn't, but we keep reviewing the network's performance).

We considered LDP Session Protection, but at the time, it 
seemed to incur quite a bit of overhead as the number of 
FEC's scaled up. However, we did see some use-cases where 
its application would be quite handy, particularly if we 
constrained the protection scheme to certain "important" 
nodes. Suffice it to say, we still haven't had a reason to 
implement it.

Cheers,

Mark.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <https://puck.nether.net/pipermail/cisco-nsp/attachments/20100623/b89ededf/attachment.bin>