Hi guys,
I was wondering if someone could shed some light and clarify the following questions:
1. When building TE LSPs with dynamic path selection, you can use TE weights instead of IGP costs. Assuming you do that and the path chosen is completely different from the path that would have been chosen if you used IGP metric, will the cost of going through the tunnel still be IGP cost from head to tail?
2. If you have two possible paths to use from head to tail, TE docs state that LSP is built via the one having lowest cost to go over it. Is that once again IGP cost regardless of the fact if TE LSPs are actually configured to use TE weights?
3. Autoroute only sets routing for LSP headend, while Forwarding Ajecency does that for every router. what if you redistribute statics on headend with autoroute enabled, will you be getting the same results as FA?
I am asking because FAs require bi-directional tunnels, which create administrative burden to implement rather than unidirectional ones sufficient for Autoroute.
4. TE docs say that "you will never loadbalance between TE path and IGP path for tunnel end". What do they mean "tunnel end" in this case? Are these all connected interfaces of a router where TE LSP terminates or is this a specific IP configured as "tunnel destination" on tunnel headend?
5. Let's say you have TE tunnel A -> B with C being "behind" B. Let's also assume that TE tunnel is configured to use IGP for path selection. In this case as far as TE is concerned the path to get to C is a path from A to B (IGP cost from A to B) and IGP cost from B to C. But that's the same as just IGP cost from A to C! Will there be loadbalancing between IGP and TE paths in this case?
It would also appear that this is a pretty much common setup, so is it true to assume that if you don't touch TE metrics on the tunnel by using "tunnel autoroute metric" you will ALWAYS loadbalance between IGP and TE paths?
Thanks,
David
