[c-nsp] MPLS Q's

[Oliver Boehmer (oboehmer)]

> Hopefully one of u will be able to answer my Q's which maybe more out
> of ignorance and/or limited experience.Any pointers and URls will
> surely help.  
> 
> 1. What exactly is the difference between "Explicit Null" and
> "Implicit Null". Does Cisco implementation use Explicit Null at all ? 

Implicit-Null tells the neighbor to pop the topmost tag. Explicit-Null
(which can be enabled in IOS using "mpls ldp explicit-null") tells the
neighbor to send the label "0".

> Some MPLS-TE Qs:
> 
> Mainly I am trying to understand the configuration options and
> behavior when I specify a path-option under TE tunnel interface.
> While creating 'ip explicit-path' the following options are
> available: Strict and loose   
> 
> While specifying the path-option under tunnel interface the following
> options are available: explicit and dynamic, both can be specified
> with lockdown option  
> 
> Now my questions are:
> 
> A. Let us say I specified two path-options, one is explicit and has
> higher preference and the other is dynamic and has lower preference.
> Initially let us say that the explicit path was not available and the
> TE tunnel was setup using the dynamic path option. After sometime if
> the explicit path becomes available, will the Head end node tear down
> the already established dynamic path and recreate the tunnel using
> the explicit path?  

Yes. By default, the head-end tries to re-optimize the tunnel ever hour
(you can tune this interval via "mpls traffic-eng reoptimize timers
frequency <seconds>") or upon certain events ("mpls traffic-eng
reoptimize events link-up"). When a "better" path is available, the
tunnel will be established along this path.  
You can also force reoptimization manually via EXEC "mpls traffic-eng
reoptimize"

> B. The same question as A but the dynamic option specified as
> 'lockdown'. In this case what will be the behavior when the explicit
> path becomes available?  

When a TE tunnel uses a "lockdown" path, it is not subject to
reoptimization (see above), unless the path goes away. 

> C. If I understand correctly the 'lockdown' option means that the LSP
> is not a candidate for re-optimization. By default it seems like all
> path options are 'candidates for re-optimization', is there any
> reason for this? Meaning, why should an already setup LSP be torn
> down (in case better path is available), will it not affect data
> traffic flow?     

It will generally not affect the traffic flow, all it takes for the
headend router to use the new path is to change an entry in the LFIB,
and this goes more or less instantly.

	oli