Tom,
To answer one that you asked a couple opf postings ago:-
I am still not clear how loadbalancing will work as bandwidth are unequal and tunnels will terminate on different routers for redundancy. The interface bandwdith on GRE tunnels does NOT refelct the actual bandwidth of the physical circuits. The bandwidth is ONLY used in dynamic routing route metric calculations. As previsouly stated a default bandwidth on a GRE tunnel is 9Kb, this does not reflect the physical circuit speed. The dynamic routing protoclcol will use the 9Kb value in route path metric calculation.
OSPF will only use this - so OSPF will see both tunnels and cost them the same.
EIGRP K values (Metric calculation) by default are Bandwidth and Delay - a default delay of a GRE tunnel is 500000 milli sec.
So with regards to your diagram, if you configure a Tunnel on RT1 @ location A and a Tunnel on RT2 @ location, A both routers will get the route from location B. If you configure 2 x Tunnels on RT1 @ location B - it will get the same both route from both tunnels - this router will auto load balance between the tunnels. Location A routers will not auto load balance. Now assuming OSPF is running over the tunnels, the device @ location A that is behind the two routers will recevie 2 routes from RT1 & RT2 - as they are of equal cost - that device will LB.
**** WARNING ****
If you run EIGRP over the tunnels and you have OSPF on the LAN @ both locations - you will have to perform mutual redistribution @ location A on RT1 & RT2 - be VERY careful, you will quite easily introduce a routing loop.
