[j-nsp] eBGP Failover

[Adam Vitkovsky]

Hi 
> Peter Ehiwe
> Sent: 30 June 2015 09:10
> Possible options :
> 1) link bundling technologies   And run single  ebgp session over the
> bundle
> 2) two ebgp sessions with shorter keep Alive
> 3) two ebgp sessions with bfd
> 
With LACP you are reliant on the mercy of the service provider.
And besides that LACP is tricky so you might not get the failover times you are after.

You could use BFD, LFM or get fancy and use SLAX script to track the availability of your static routes next hops today.
I guess the BFD is the easiest one:
set static route x.x.x.x/30 bfd-liveness-detection minimum-interval 50

I would not recommend doing two eBGP sessions as it consumes twice the amount of memory to keep RIB-in for two neighbours (default on Junos).


adam