Marked for the PM to review and thanks for the feedback.
Good idea. For our sites with bonded connections we just monitor the interface status, which is normally sufficient, but this would not detect the lack of traffic on a primary/backup link (i.e. remote site->internet routing issues would not trigger alerts) - it would only find that a given interface is not up. I suppose what would be considered useful depends on the network topology though.
To make coding this easier, I'd suggest just giving a node the ability to have multiple IPs and then giving each IP a label. If one goes down, put it in to a "DEGRADED" state, rather than "IN BACKUP", because perhaps the backup link is down rather than the primary, or the node might have several bonded connections.
One instance where I would find this useful is by giving a node both internal and external (internet) IP addresses - this would mean that I could get an alert indicating that the external IP is fine, but the VPN device is not connecting, which means that only the VPN device needs rebooting first rather than the briged DSL modem. Of course, perhaps it would be a challenge to find the Internet-facing IP of all devices automatically :)