The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2001-Jun> msg00342



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

Label Mapping Processing !!

  • From: Eric Gray <eric.gray@sandburst.com>
  • Date: Fri, 22 Jun 2001 10:35:33 -0400
  • Cc: mpls@UU.NET

Manoj,

    So, if you look at the test in LMp.28, you see that LMp.29 is skipped if
there are no outstanding label requests.

    The behavior you're pointing out is correct as is.

    If the LSR is using either independent or ordered control, downstream
unsolicited label distribution, the only way that it would have outstanding
label requests is if the upstream LSR needs more than one label and has
not yet received additional ones it asked for.  The idea is that - in DU mode
- the upstream LSR will get one label for free; if it asks for a label, it must
be because it wants at least two.  Therefore, sending labels for outstanding
requests is definitely not redundant.

--
Eric Gray

You wrote:

> Hi,
>
> I have a few doubts in step LMp.29 Section A.1.1 (RFC 3036). i thnk the
> check for Downstream Unsolicited
> Independent/Ordered Control is redndant. In case of DU independent
> control,  a mapping is sent to peers as
> soon as a new fec is recognized and DU ordered control is handled by
> LMp.19-LMp.21.   For pending
> label requests,,step 1-5 given for DoD should be sufficient for all
> posible cases.  Let me know if this is correct.
> Also,  why is stored hop count value set to UnknownHopCount (0), in this
> step ?
>
> thanks in advance
> manoj