[c-nsp] Loopback Interfaces and VRF configuration

[Phil Mayers]

Maya Yajnik wrote:
> Hello
> 
> Does anyone know the purpose of using an additional Loopback interface (like
> Loopback1) and associating that with a vrf? I see this in VRF-Lite example
> configurations. 

Lots of reasons. In sort-of order of likelihood:

  1. You might be running OSPF inside the VRF, and best practice is to 
have the router ID be an IP on a loopback interface. It's *still* best 
practice when VRFs are deployed.

  2. You might be running an IGP and iBGP inside the VRFs, and using 
iBGP for the peering to adjacent non-VRF aware routers

  3. You might be putting your netops inside a VRF and want to 
telnet/SSH to the loopbacks without having to leave the VRF.

  4. You might be talking to NTP/Syslog/radius/tftp/other servers on a 
per-VRF basis and want to use a consistent source IP.

  5. You might have the internet in a VRF (and customer or yourself in 
"default" vrf) and want to source GRE/IPSec tunnels on the internet (now 
VRFed) side from consistent interfaces.

And so on