Re: [Pce] Working Group Last Call:draft-ietf-pce-pcep-svec-list-02.txt

["Itaru Nishioka" <i-nishioka at cb.jp.nec.com>]

Hi Filippo

Thank you for comments.

For independent request, it is not required to reply all combinations (just 3 EROs replied in
section 6.2), because they are independent path computation.For dependent request, let me clarify
your points.
- For dependent (associated) diverse path requests in a single domain, concurrent path computation
are required to meet good resource usage or maximize disjointness between diverse path pair.
- When this concept is applied across PCE domains,  downstream PCE must reply full combinations of
potential path to meet same objective above,  because final path decision would be made in upstream
PCE using BRPC procedure.
The second point was a reason why dependent diverse path request replied so much EROs (9 EROs).
Right?

If clarification above is correct, the same solution as in the independent case can be applied. This
is because downstream PCE have responsibility on optimization of own domain, not upstream PCE. So
the downstream PCE can choose suitable EROs set (just 3 EROs in section 6.2 example) based on own
policy.

Other typos will be revised in next version.

Thank you.
== Itaru Nishioka,

 

> -----Original Message-----
> From: pce-bounces at ietf.org [mailto:pce-bounces at ietf.org] On 
> Behalf Of Filippo Cugini
> Sent: Thursday, September 03, 2009 1:47 AM
> To: pce at ietf.org
> Subject: Re: [Pce] Working Group Last 
> Call:draft-ietf-pce-pcep-svec-list-02.txt
> 
> Hi,
> 
> just one clarification in the case of multiple associated 
> end-to-end diverse path computations.
> Example: consider one additional end-to-end diverse path 
> computation (say Requests #3 and #4, source S2 and dest. D2) 
> associated to the diverse path computation S1-D1 reported in 
> section 6.
> The EROs computed for Req #1 and #2 (those in section 6.2) 
> may be related to the EROs computed for Req #3 and #4.
> Thus, up to 9 EROs per request could be necessary in each path list.
> However, if the two sets of computed EROs are independent, up 
> to 3 EROs per request would be sufficient (as in 6.2).
> Is it required to enclose in the PCRep message all the 
> combinations (even if they are independent)?
> 
> Few typos:
> - section 4.1, "may in the SVEC may"
> - 6.2, last line, VSPT2 -> VSPT3
> - 6.3, VSTP -> VSPT
> 
> Br,
>   Filippo
> 
> 
> 
> ----- Original Message -----
> From: "JP Vasseur" <jvasseur at cisco.com>
> To: <pce at ietf.org>
> Sent: Tuesday, September 01, 2009 8:30 PM
> Subject: [Pce] Working Group Last Call: 
> draft-ietf-pce-pcep-svec-list-02.txt
> 
> 
> > Dear all,
> >
> > This starts a 2-week WG Last Call for draft-ietf-pce-pcep-svec-
> > list-02.txt, which has been stable for quite some time, and 
> is ready  for
> > WG Last call. Please send your comments by September 16, 
> 9:00am ET  on the
> > mailing list.
> >
> > We would also welcome feed-back from potential implementers.
> >
> > Thanks.
> >
> > JP and Julien.
> > _______________________________________________
> > Pce mailing list
> > Pce at ietf.org
> > https://www.ietf.org/mailman/listinfo/pce
> 
> _______________________________________________
> Pce mailing list
> Pce at ietf.org
> https://www.ietf.org/mailman/listinfo/pce
>