[c-nsp] Reasons for "random" ISIS flapping?

[Pete Lumbis]

I don't know anything about Juniper, but in general LFA is just a step
ahead at the protocol level. We do a second SPF run on the remainder of the
routes after picking the best and install all those as backup paths. We
still need something to trigger the switchover. The only thing I can think
of is how you were simulating failure. Lost of carrier will always beat
BFD.


On Thu, Aug 29, 2013 at 10:41 PM, Mark Tinka <mark.tinka at seacom.mu> wrote:

> On Thursday, August 29, 2013 06:00:05 PM Pete Lumbis wrote:
>
> > I don't see it as an either/or question. You still need
> > BFD for failure detection. Nothing happens until a
> > failure is detected. (r)LFA fixes the rest of the
> > convergence equation, mainly the time it takes to notify
> > neighbors and recompute a path and what can be the
> > biggest delay, re-programming the data plane on hardware
> > platforms.
>
> The reason I'm thinking about this is because I recently
> lab'ed up some Juniper kit, and didn't notice any real
> difference in failure detection and local re-routing with
> BFD on or off, when the box had LFA running.
>
> Of course, real life at a larger scale is a completely
> separate issue, but I'll start to collect some hard data on
> this and see if (assuming reasonably good LFA backup prefix
> coverage) gains are positive, marginal or even negative.
>
> Mark.
>