Re: TE Requirements Draft - ELSP

[Jim Boyle <jboyle@pdnets.com>]

Francois,

Let's say I have a decent idea of the mix between Voice-Signalling and
Voice-Encapsulated, say for example that it is 1:4.  Also, say that a
router it can distinguish Voice-Signalling and Voice-Encapsulated and
Other.  I may wish to only use one LSP for Voice-Signalling and
Voice-Encapsulated to a particular destination, and another for Other.
I've reduced the number of LSPs used from 3 to 2.  If there are many
Other, then the reduction can similarly be 2:1.  As for queue
selection, etc..  As I said, the router has the ability to
differentiate the different types of traffic.  On ecapsulation to the
LSP, the appropriate "Experimental" bits are set.  These affect queue
selection along the path.

So - why you ask might one want to use only the number of LSPs absolutely
necessary - I think that's obvious - signalling scalability (yes,
potentially trading off on phb precision - oh well).  So, in this case,
two OAs become one class, which route against one class-type, one cspf.

Hope that answers your question.

Jim


On Mon, 19 Nov 2001, Francois Le Faucheur wrote:

> At 13:38 16/11/2001 -0500, Jim Boyle wrote:
>
> >Francois - if your question is:
> >
> >-----------------------
> >Would service providers wish to potentially send more than one, OA down an
> >E-LSP, which makes use of class-type specific routing information?
> >-----------------------
> >
> >The answer is yes.
> >
> >Picking on poor voice, one OA might be signalling traffic, one might be
> >VoIP data.  They might go into different queues.  They might not need to
> >be on seperate LSPs.
>
> I am still struggling to picture a whole deployment scenario:
> in this example would you assume that :
>          (i) a path has been computed separately for voice sig and voice
> flows (each with a separate bw requirements and potential differnet
> bandwidth constrainta/CT). In that case, I feel that it woudl be simpler to
> just put those on differnet LSPs. no?
>          (ii) a path has been computed for one of the two , say voice-flows
> (based on the voice-flow bandwidth requirements and using the voice-flow
> bandwidth constraint/CT). In that case, why woudl you want to send the
> voice sig on that tunnel? why not sent it on its SPF? sending voice-sig on
> the tunnel routed for voice flow will just result in potentially sending on
> a longer-than-SPF path which you have no idea whether it is actually better
> or worse than the SPF-path considreing teh voice-sig resources (which are
> independent from teh voice flow resources). In other words, if I only do
> CSPF for one OA, I woudl personnaly be inclined to send the other OA on an
> SPF-LSP rather than on that CSPF LSP because it has just as many chances of
> making it worse than making it better. no?
>
> Thanks
>
> Francois
>
>
> >The LSP might draw upon resources of a non-default
> >class-type.
> >
> >Hope that's not too fancy ;)
> >
> >Jim
> >
> >On Fri, 16 Nov 2001, Francois Le Faucheur wrote:
> >
> > > At 10:04 16/11/2001 -0500, Nabil Seddigh wrote:
> > > >I thought I should get this comment in before the -02 version
> > > >of the Requirements draft emerges.
> > >
> > > just in time...
> > >
> > > >Hopefully the authors will
> > > >consider the following suggestion:
> > > >
> > > >- I think it is useful to put an explicit requirement that the TE
> > > >    solutions need to support both E-LSP and L-LSP. The way the
> > > >    draft reads at the moment, it does not come through clearly.
> > > >    Most of the examples and wording would lead one to believe that
> > > >    the DS-TE solutions should only focus on L-LSP.
> > >
> > > That's because the requirements indeed assumes that DSTE will be deployed
> > > using L-LSPs (or E-LSPs but on which traffic from a single OA is carried).
> > > soon-to-be-released-02 already has text clarifying this.
> > > When deploying DS-TE, the SPs I spoke to wanted not only to do per-class
> > > admission control but also per-class routing, which normally requires use
> > > of L-LSPs (or E-LSPs transporting a single OA).
> > >
> > > Is allowing operations of DS-TE over E-LSPs which transport a single OA
> > > what you're after (in which case it's already there)?
> > >
> > > Or are you suggesting operations of DS-TE over E-LSPs which transport
> > > multiple OAs?
> > > In that case, what are the Service Provider requesting this?
> > >
> > > Could these SPs clarify whether:
> > >          -  they would use this to do per-class admission control but not
> > > do per-class routing (would it not be a pity to deploy all this
> > > sophistication and have voice not being able to take its shortest path
> > > simply because there is no more resources for data on a link)?
> > >          -  they expect the routers to do fancy things like
> > > dynamic/on-the-fly mapping of OAs on various LSPs : compute path for
> > > differnet OAs separately and then if they happen to be the same at one
> > > point in time push them onto the same LSP? then what happens if later on,
> > > one OA gets preempted and not the other one: remap one OA on another LSP
> > > with some other OAs?
> > >
> > > Thanks
> > >
> > > Francois
> > >
> > >
> > > >Best
> > > >Nabil Seddigh
> > > >nseddigh@tropicnetworks.com
> > >
> > > _________________________________________________________
> > > Francois Le Faucheur
> > > Development Engineer, IOS Layer 3 Services
> > > Cisco Systems
> > > Office Phone:          +33 4 97 23 26 19
> > > Mobile :               +33 6 89 108 159
> > > Fax:                   +33 4 97 23 26 26
> > > Email:                 flefauch@cisco.com
> > > _________________________________________________________
> > > Cisco Systems
> > > Domaine Green Side
> > > 400, Avenue de Roumanille
> > > 06 410  Biot - Sophia Antipolis
> > > FRANCE
> > > _________________________________________________________
> > >
> > >
>
> _________________________________________________________
> Francois Le Faucheur
> Development Engineer, IOS Layer 3 Services
> Cisco Systems
> Office Phone:          +33 4 97 23 26 19
> Mobile :               +33 6 89 108 159
> Fax:                   +33 4 97 23 26 26
> Email:                 flefauch@cisco.com
> _________________________________________________________
> Cisco Systems
> Domaine Green Side
> 400, Avenue de Roumanille
> 06 410  Biot - Sophia Antipolis
> FRANCE
> _________________________________________________________
>