The MPLS WG Archive

[Jean Philippe Vasseur <jvasseur@cisco.com>]

Hi,

At 19:12 19/11/2001 -0500, Sachin Kalra wrote:
>Hi:
>
>I have few questions about "Bypass Tunnels". Any help is appreciated.
>
>In following figure Primary LSP traverses from (R1->R2->R3->R4->R5) and 
>Bypass tunnel from (R2->R6->R7->R8->-R4). Bypass is denoted by double lines.
>
>
>         [R0]
>              \
>[R1]---------[R2]------------------[R3]-----------------[R4]-----------------[R5] 
>
>                   \\                                        //     \
>                    \\                     //    \
>                 [R6]======[R7]======[R8]        [R9]
>
>
>If R4 is using Interface specific Label Space. (i.e. it does not use 
>Global Label Space). Then we need to signal LSP through "Bypass Tunnel" 
>path to obtain label assigned out of the label space used by R4 on the 
>interface terminating the link from R8 (Which will later be used by R2 for 
>label swapping before stacking the label for Bypass Tunnel)
>
>In this case my understanding is that once R2 has the "required label" 
>from R4, then R6, R7, R8 do not need to maintain any state for the 
>signaled LSP, nor do they need any NHLFEs for that LSP. Because traffic 
>will be sent through the "Bypass Tunnel".
>
>1. Is this correct?

yes

>2. How do we handle this scenario?

see draft-manyfolks-rsvp-fastreroute-00.txt (very recently published)

>3. Does LSPs signaled through Bypass Tunnel need NHLFEs on transit nodes?

no.

JP.

>Thanks for your time,
>Sachin Kalra