[j-nsp] MPLS Question

[Phill Jolliffe]

See in line tobelow

On Mon, Apr 27, 2009 at 6:00 PM, William Jackson <wjackson at sapphire.gi> wrote:

> On my ingress router I am setting up some RSVP label paths with strict
> ERO's, I add the fast reroute option and the adaptive option.  So that
> the primary and secondary paths are already up.

Adaptive is to make the reservation style SE not FF.

Fast reroute is not about making a secondary path on the ingress. You
can generally think of fast reroute as a mechanism to help protect
traffic for the period it takes to switch from the failing primary
path to a secondary path. When the secondary path does not have the
standby keyword to cause it to be pre signaled this has more value
again.
>
> On my transit nodes I add the RSVP link-protection command to the
> relevant interfaces.
>

If your turning this on in proto rsvp then you want to use the
link-protection keyword under your LSP and not the fast-reroute option
you mention above. You are mixing to different forms of FRR otherwise.


> On my ingress node I also see options for node-link-protection and
> link-protection under the label-switched-path section, are these all
> needed on the ingress node to get fast-reroute working
>

see above, imho node protection is usually over kill. link is fine.

>
> I am getting a single ping packet lost when I pull a cable out was
> wondering if there was a way to get even faster path swap over?

I guess you see the packet lost while running a ping rapid or similar?
Test results here can be misleading due to the way your ping app is
coded. Does it send another packet before receiving icmp echo reply to
the last? and if not how long does it wait.

Unless you have an excessive number of LSP on the node connected to
the failing link I would guess you could measure the period of packet
loss being sub 50ms (if you had a decent test program/traffic
generator). FRR will minimize packet loss but never completely
eliminate it.

>
> Do I have to tweak any timers or something?
>
>
>
> Thanks
>
> _______________________________________________
> juniper-nsp mailing list juniper-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/juniper-nsp
>



-- 
Phill Jolliffe