Re: [Idr] I-D Action: draft-ietf-idr-add-paths-10.txt

[Jeffrey Haas <[e-mail]>]

On Fri, Oct 24, 2014 at 04:57:57PM +0000, Alvaro Retana (aretana) wrote:
> I had put an implementation report in my To Do list, but didn?t get to it this time..  If you are ok with it, I can work on it and present it in Hawaii.  I know that would violate the ?no draft, no presentation? slot rule, but it would let us move forward.  I can present an update on the latest changes to close the loop as well.

And if the survey contents were ready in the next week, I promise to have a
report from Juniper to contribute prior to Hawaii.

> BTW,  we also updated the 'BGP Persistent Route Oscillation Solutions? draft (draft-walton-bgp-route-oscillation-stop-09): it documents how ADD-PATH can solve the MED churn, which is that started this whole thing.  We would like to ask for WG adoption.

I'll add this to my review list.  I agree with Alvaro that the scope of the
changes requested in the document make it appropriate for IDR adoption.

To add a dose of operational input on the draft, I've more typically seen
customers asking for add-path with regard to traffic-engineering
applications than for oscillation damping.  There are two ugly challenges
add-path solely for the oscillation case introduces:

- You have to have it turned on to stop the issues.
- The additional state in your network, when you don't need it, consumes
  resources that otherwise may not be needed.

What's unfortunate is the infrastructure to depend this sort of persistent
oscillation is itself expensive making it ugly to only turn on group-best
when necessary.

No answers, only comments.

-- Jeff