The MPLS WG Archive

[Peter Willis <pjw@ip-engineering.bt.com>]

> but mis-routing is a rare case even in MPLS networks.  You say you need
> to be able to test for this.
> 
> but you say that there is no need to test for rarer cases.
> 
> I think we need some way to quantify how rare these cases (FIB
> corruption and FIB corruption where the egress interface is null or is
> an interface that will result in traffic being forwarded back to the
> router with the corrupted entry) are?  It is non-obvious to me that the
> first needs detection whilst the second doesn't.

The issue is not to do with the frequency of the fault but the impact of the 
fault:

In case where we use IP in IP tunnels if there is a forwarding error then the 
tunnel packets will normally go into a routing loop so the VPN IP packets do 
not get delivered to the wrong customer so there is no security problem.

In the case of MPLS if there is a forwarding error the VPN IP packets can get 
delivered to the wrong customer so there is a security problem.

So for IP VPNs using IP in IP tunnels we can live without detection of 
mis-routing (although a tunnel livelness indicator would still be useful).
But for IP VPNs using MPLS we need something to detect mis-routing.

Peter.