Dogsbod,
I have similar network thoughts... larger scale, but same idea.
At my main office, I have several 3550 POE switches stacked together and uplinked into a gigabit switch for my server farm and fibre farm.
i have another campus connected to this via a 100meg Point to point fibre link (i use that for disaster recovery).
My "core" VPN router is a 1710
My remote sites have a 1760V with Wic0 being a 10BT ethernet wic, VIC 0 and VIC 1 are 2FXO for my call manager implementation. I also have the 3DES VPN Module on the 1760V and have tunnels built between my 1760V (remote) adn 1710 "central"
For backup, I have a business DSL connection. I route all my internet traffic out of IPCop (
which has a nifty open source VPN connection that I use to build a redundant tunnel to another 1710 at central.
I don't have "automated" failover. I have to manually switch the routes both at central and remotely. The good part is, IPCop does port fowarding too, so if primary link fails, I can still "backdoor" into my remote router to change routes using IPCop/VPN
7 remote sites, 30 servers, 500 workstations, 150 phones, Call Manager, Unity, 10TB of active data and 99.94% uptime the last 18 months since I started.
That's "cheap" redundancy...
--DW