Reading through O'Reilly's Network Warrior in the chapter "Ethernet Resiliency he describes the following Scenario (Trying to keep this in the limits of fair use):
"Assume that the link between New York and Los Angeles suffers a partial outage. Something has happened to cause the serial interface on NY-Primary to enter a state of up/down, but the serial interface on LA-Primary has stayed up/up. I’ve seen this more than once on different kinds of circuits.
HSRP responds to the down interface on the New York side by making the NY- Backup router active because we’re tracking the serial interface on NY-Primary. Packets are forwarded to NY-Backup, and then across the T1 to LA-Backup, which forwards them to their destinations. The return packets have a problem, though. As the LA-Primary router does not recognize the link failure on the primary T1, it has remained the active router. "
It then goes on to say:
"A more robust solution to a link-failover scenario is to incorporate an interior gateway protocol running on all of the routers. A protocol like OSPF or EIGRP establishes neighbor adjacencies across links. When a link fails, the routing protocol knows that the remote neighbor is unavailable, and removes the link from the routing table."
My Situation, and Question:
This seems like what I might be looking for. I will have to links from Datacenter provider and I have to routers I would like to have be redundant. HSRP can help me on the LAN side, but since their routers can't talk to mine, HSRP from them won't help with WAN redundancy because if one of routers fails the provider's won't know about it. Is EIGRP a possible solution in my scenario like the one described (Assuming a EIGRP ASN can be shared between my router and the datacenter, this normal?) ? Can anyone explain how this works with an example, or have a link to something I can read that shows examples of how this works?
In my case, the provider will own the address and not myself, so they advertise them on the internet and have redundant providers for the ip addresses. Also, my goal is failover, not balancing.
Here are some related questions on my recent exploration of network redundancy: