The MPLS WG Archive

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

Hi Snigdho,

At 11:04 27/02/2002 -0800, Snigdho Bardalai wrote:
>Hi,
>
>Could you please help me understand how the NNHOP
>Bypass tunnel works as described in Secion 6.5.
>
>Consider the network below (from Section 1.2):
>
>                 [R8]
>                     \
>               [R1]---[R2]----[R3]----[R4]---[R5]
>                          \\          //   \
>                           [R6]===[R7]     [R9]
>
>Say, labels available at R2 are L1 for the working
>tunnel from R3 and L2 for the bypass tunnel from R6.
>
>Again, labels available at R3 are L3 for the working
>tunnel from R4.
>
>Now when a packet is switched from the working tunnel
>onto the bypass tunnel at R2 the label stack (as
>described) will be L2->L1.

No. In this case, the bypass tunnel is going to the NNHOP (R4 in the 
example above) so the PLR (R2) will redirect the traffic onto the BYPASS 
tunnel, replacing L1 by L3, pushing L2 and redirecting the traffic to R6 
(label stack = L2-L3). As described in section 6.1 of the draft, R2 learns 
the backup label to use (L3) from the RRO (prior to the failure). Then R7 
will perform a PHP and the MP (R4) will receive L3 as expected (but from a 
different interface). With a BYPASS tunnel going to the NHOP, the PLR would 
have acted as you described above. This applies to global label space at 
the Merge point.

JP.

>When the packet leaves R7 (the penultimate hop) the
>label stack is L1.
>
>At R4 it is expecting to L3, whereas it sees L1, won't
>that be an error ?
>
>Would appreciate a reply,
>Thanks,
>Snigdho
>
>__________________________________________________
>Do You Yahoo!?
>Yahoo! Greetings - Send FREE e-cards for every occasion!
>http://greetings.yahoo.com