The MPLS WG Archive

[Ping Pan <pingpan@juniper.net>]

Doug Degan wrote:

> in section 4.3  it is said that the PLR should try to find the bypass 
> tunnel. NHOP/NNHOP one.
> 
>  
> 
>  my question is what are the assumptions that are made when saying that 
> the PLR is capable to do so?
> 
>  
> 
> 1.a) The PLR should choose from the Bypass tunnels available, meaning 
> tunnels which this router is their Ingress. 


Yes.

> should it know their 
> explicit route, or their recorded route is enough?
>


The bypass LSPs can be established independent from the protected LSPs. 
When protecting a LSP, a PLR just uses a bypass LSP as an alternative 
route for the protected LSP. Knowing ERO or RRO is an implementation issue.


>  
> 
> 1.b) It should parse the current (protected) LSP's recorded route. does 
> it gives enough information?
> 
> the PLR needs to know which node/link to avoid and what bypass 
> destinations are good.
> 
> as far as I know - the recorded route doesn't have to include router's 
> ID/addresses.
>


Huh? RRO, Subtype = 1.


>  
> 
> 2.a) what happens if TE is not enabled in the PLR or in another node 
> along the protected lsp's path - does it mean that FRR wil not work?
> 
>  
> 
> 2.b) what if TE is enabled anywhere but CSPF is not used (and no 
> explicit path is found in the candidate for bypass tunnels)
> 
> 


The most important task for bypass LSPs is to avoid some links or nodes. 
   If this cannot be doe intelligently, I would recommend to use 
whatever the means available to get the job done.... (type the path by 
hand, for example ...: -))

- Ping