[c-nsp] Alternate setup for "better" HA

[John Elliot]

> 
> If the 2851s have independent connections to your core, you can connect 
> the second 2851 to the first 2960 (and the first 2851 to the second 
> 2960), create the SVI interfaces with the same IP address as the first 
> 2851, and keep them in shutdown.

2851's will connect to the "rest" of our network via (currently) single QinQ hand-off from a carrier, that is connected to 2960.
The second 2851 we can reach via OOB (So can unshut the portchan(physical gig int)) and it would have identical config bgp/vrf's/ospf etc(In current setup)), so time to recover in this setup is not ideal, but could be worse!
> 
> Thus if your first 2851 fails, you can telnet to the second and manually 
> "no shut" them: not a big deal, but usually faster than running at the 
> POP. You can also make a clever HSRP + EEM setup to have this 
> automatically done, but I personally would not use such a complicated 
> setup for such a minor achievement.
Yep, I considered EEM - played with a few scenarios(x-connect between the two  2851's, and the second 2851 simple running ip sla to test is first 2851 is reachable, if it is, trunk port to 2960 stays down, if not, it brings trunk port up with EEM)....worked wel(apart from arp issues)l, but potential for both 2851's to have trunks to 2960 up.
> 
> For the 2960s, if your carrier gives you only one physical connector 
> there is not much that you can do to achieve redundancy: the connector 
> has to be plugged in some device (here it's the 2960), and if it fails 
> you have to move it.

Yep...not much I can do here unfortunately....it's looking like the most basic solution is still the manual config replication, and manual unshut of trunk interface....
hehe wish Cisco would bring out a "stackable" router option like they have with switches.