The MPLS WG Archive

[Francois Lemarchand <Francois.Lemarchand@cosinecom.com>]

Title: RE: Decreasing TE-LSP bandwidth, "interoperability issue" (fwd)

Ok,

Then it's maybe not a technical argument as such
but a best practice to reduce the number of special
cases : changing the LSP-ID for bandwidth increase/
decrease regardless of a path change.

Although it's not optimal it didn't prevent using FF
style. So I didn't see a contradiction by generalising
this LSP-ID change while keeping FF as one of the
reservation style.

Francois

-----Original Message-----
From: Jim Boyle [mailto:jboyle@pdnets.com]
Sent: mardi 5 novembre 2002 14:33
To: Francois Lemarchand
Cc: mpls@UU.NET
Subject: RE: Decreasing TE-LSP bandwidth, "interoperability issue" (fwd)

Yes, we all agree that as the spec states, when the path changes the LSP
ID should be changed.

The issue is that one implementation has issues if the LSP ID of
a transit LSP does not change when the bandwidth decreases.  Alternatively
you could state that the issue is that one (or two) implementations are
unique in that they don't change the LSP ID when they decrease the
bandwidth.

So the options at this point are:

o) address the implementation with transit issue when bandwidth decreases
and LSP ID is not changed.

o) require LSP ID change during BW decrease in the spec and address the
implementation(s) which currently don't do this.

We've had a few people weigh in on both sides of this, so I'm not sure
if there is any resolution.

Jim

On Mon, 4 Nov 2002, Francois Lemarchand wrote:

> Hello !
>
> For re-optimisation I guess. If there is a need to change
> the path when bandwidth increase then it's probably usefull
> to return to the old path when bandwidth decrease (probably
> more optimal).
>
> So changing path probably justify make-before-break and
> LSP-ID change.
>
> Did I miss somthing ?
>
> Francois
>
> -----Original Message-----
> From: Kireeti Kompella [mailto:kireeti@juniper.net]
> Sent: lundi 4 novembre 2002 20:27
> To: mpls@UU.NET
> Subject: Re: Decreasing TE-LSP bandwidth, "interoperability issue" (fwd)
>
>
>
> Hi All,
>
> I'd like to re-iterate Der-Hwa's question:
>
> On Thu, 31 Oct 2002, Der-Hwa Gan wrote:
>
> > It is perhaps more appropriate to ask what is the technical reason that a
> > new ID is required to get bandwidth decrease?
>
> Is there a technical reason that someone can articulate?  Along the
> same lines, is make-before-break needed on bandwidth *decrease*?
>
> Thanks,
> Kireeti.
>