RE: Valid Race Condition related to draft "draft-ietf-mpls-crlsp -modify-03.txt"

["Don Fedyk" <[e-mail]>]

Manoj

This thread has gone full circle. Just because the draft is 
most useful for CR strictly routed LSPs does not mean it is 
harmful to loose LSPs. 
If it were up to me I would put a question any wide spread use of
loose LSPs. These LSPs are operational headaches. The more 
you add the more provisioning you have to change should 
your network change. One of the benefits of the modify draft 
was to make CR strictly routed LSPs less of a an operational 
headache and an even better option than loose LSPs.
But the question is the draft only applicable to Strict LSPs?
My answer is no in fact if you wanted to modify the explicit 
part of a strictly LSP this feature would work and must work without
any race conditions (which was the original objection). 

Any implementation of the modify will have a knob to turn on 
or off the modify so if you don't want the feature don't use it.

Don




> -----Original Message-----
> From: manoj juneja [mailto:[e-mail]]
> Sent: Wednesday, April 25, 2001 6:33 PM
> To: [e-mail]; [e-mail]
> Cc: Fedyk, Don [BL60:2868:EXCH]; [e-mail]; [e-mail]
> Subject: Re: Valid Race Condition related to draft
> "draft-ietf-mpls-crlsp -modify-03.txt"
> 
> 
> Hi Don/Loa/Prem,
> 
>        Today we got the mail of making the draft 
> "draft-ietf-mpls-crlsp 
> -modify-03.txt" as a proposed standard. Should not you people 
> feel that the 
> restriction that modification of CR-LSP is only applicable to 
> strictly 
> routed hops should be addressed in the draft prior to making 
> it as a RFC. I 
> think this is a major restriction and should be addressed in 
> the scope of 
> the document.
> 
> Regards,
> manoj.
> 
> 
> 
> 
> >From: "Loa Andersson" <[e-mail]>
> >To: "Sharma Prem" <[e-mail]>
> >CC: 'Don Fedyk' <[e-mail]>,  manoj juneja 
> ><[e-mail]>, [e-mail]
> >Subject: Re: Valid Race Condition related to draft 
> "draft-ietf-mpls-crlsp 
> >-modify-03.txt"
> >Date: Mon, 23 Apr 2001 13:05:32 +0200
> >
> >All,
> >
> >some comments to this discussion.
> >
> >I agree with Perm, if the draft is only applicable to LSPs 
> where every
> >node is explicitly routed, that needs to be explicitly state early in
> >the draft. Also know from some offsets communication that "the
> >explicit routed only" approach  is under some debate.
> >
> >Given that we can have LSPs where other constraints are important, I
> >would
> >be interested in combining dynamic routed CR-LSPs (routed on other
> >constraints than explicit nodes), with one explicitly routed node,
> >e.g. for performance measurements. I would also be interested to be
> >able to make those  measurements at different places, a reason for
> >changing (modifying) the LSP, drop one explicitly routed node and
> >replace with another.
> >
> >I'm not concerned over have the loosely routed part of an LSP staying
> >alive of the strictly routed part goes away :-), but given that I can
> >retry if the a node in the loosely routed part of the LSP NAK a
> >request,
> >but not if an explicitly routed node does it. Need to distinguish
> >between
> >cases. Of course this is not a problem if explicit routed only.
> >
> >It is not only CR-LDP that sets up CR-LSPs, it is much more common to
> >do so by RSVP-TE, hence my reflection on the need to something for
> >RSVP-TE.
> >
> >/Loa
> >
> >
> >
> >--
> >Loa Andersson
> >Chief Architect,
> >Utfors Research, Architecture and Future Lab (URAX)
> >Utfors AB
> >Råsundavägen 12
> >Box 525, 169 29 Solna
> >Office          [phone]
> >Office direct   [phone]
> >Mobile          [phone]
> >Email           [e-mail]
> >WWW             www.utfors.se
> 
> _________________________________________________________________
> Get your FREE download of MSN Explorer at http://explorer.msn.com
> 
> 
>