The MPLS WG Archive

["Don Fedyk" <dwfedyk@nortelnetworks.com>]

Title: RE: Any more comments on draft-ietf-mpls-te-feed-00.txt?

Dear Atsushi Iwata,

The feedback draft is aimed at numbered links only. The cycle is
a path is selected from a topology database that is populated by
an IGP. Both IGPs use the same format so it does not matter which
IGP. The path becomes the ER MPLS LSP.

Next the ER LSP is established and our Draft uses CR-LDP to accomplish
this. If a link that is not able to support the path is encountered,
the bandwidth profile (independent of the IGP for the number link) is
added to the failure notification. This information is generated for
every link along the traversed section of the path.

This information can then be added to the topology database improving its
accuracy for all subsequent path requests. 

A crankback as you describe cannot update the database since it may
not have an accurate bandwidth profile. This could happen if you
encountered a node that did not support the feedback. So crankback
could be the degenerate case of feedback.

ER paths can support un-numbered links in some cases. When the path is
loose routed, for example, but this is not the constraint based case
we were covering in the draft.

Don

> -----Original Message-----
> From: Atsushi Iwata [mailto:iwata@ccm.CL.nec.co.jp]
> Sent: Wednesday, April 05, 2000 8:30 PM
> To: Ashwood-Smith, Peter [CAR:CS57:EXCH]
> Cc: MPLS mailing list; Norihito Fujita
> Subject: Re: Any more comments on draft-ietf-mpls-te-feed-00.txt?
>
>
> Hi, Peter,
>
> I have one comment.
>
> At 10:17 00/04/05 -0400, Peter Ashwood-Smith wrote:
> >RE: draft-ietf-mpls-te-feed-00
> >
> >   Still looking for feedback on the feedback draft ... pun
> >intended.
> >
> >   I will be asking the WG chairs for a last call on this in the
> >next few days unless somebody has some issues with it.
>
> As I explained our draft,
> draft-fujita-mpls-crldp-crankback-00.txt, in the
> last IETF, there are several possibilities of indicating of
> links or nodes.
> If you use unnumbered point-to-point links between routers,
> you cannot
> specify IPv4 address of interfaces for IPV4 specificed link
> feedback TLV
> and IPV6 specified link feedback TLV. In that case, you have to use a
> routing protocol (OSPF etc) dependent TLV to indicate links or nodes.
>
> Our draft specifies protocol independent TLV (using ER-HOPS
> information)
> and routing protocol dependent TLV (OSPF, ISIS-specific
> information). I
> guess that our draft could be an one of possible solution to
> indicate such
> links or nodes.
>
> What do you think ?
>
> Best regards,
> ----------------------------------------------------------------
> Atsushi Iwata
> Assistant Manager
> Network Architecture TG, Networking Research Laboratory
> C&C Media Research Laboratories
> NEC Corporation
> 4-1-1 Miyazaki Miyamae-ku, Kawasaki, Japan 216-8555
> TEL: +81-44-856-2123 (Direct), Fax: +81-44-856-2230 (Direct)
> NEC-internal TEL: 8-272-3281, NEC-internal FAX: 8-272-3299
> E-mail: a-iwata@ah.jp.nec.com
>
> ** New organization has started since Apr.1, 2000. ***
>
>