RE: could we agree on this ? Fwd: RE: TE Requirements Draft - ELSP

[Shahram Davari <Shahram_Davari@pmc-sierra.com>]

Francois,

I think i,ii, iii(a), iv should be supported. No need to support iii(b), unless there is real demand.

Yours,
_Shahram 

> -----Original Message-----
> From: Francois Le Faucheur [mailto:flefauch@europe.cisco.com]
> Sent: Monday, December 03, 2001 9:14 AM
> To: Nabil Seddigh
> Cc: Francois Le Faucheur; Shahram Davari; te-wg@ops.ietf.org
> Subject: Re: could we agree on this ? Fwd: RE: TE Requirements Draft -
> ELSP
> 
> 
> Nabil and all,
> 
> >In this regard, I would suggest that your options are actually:
> >
> >   i) Using E-LSPs with traffic from a single OA
> >  ii) Using E-LSPs with traffic from multiple OAs with single BW
> >  iii) Using E-LSPs with traffic from multiple OAs with multiple BW
> >   iv) Using L-LSP
> >
> >There seems to be prior agreement that options (i) and (iv) are
> >required.
> 
> So let's record that we can agree on the proposal to update 
> REQTS draft to 
> allow support for (i), unless we hear other views.
> (iv) is already there.
> 
> >The open questions seem to be whether or not (ii)
> 
> We need to be more specific on (ii). I believe we have 
> established that:
>          - (ii) can be supported without any protocol 
> extensions. Whether 
> we put it in the REQTS document or not, it will not affect the DSTE 
> protocol extensions and it can be implemented in a head-end 
> LSR. As I said 
> earlier, it comes for free from a "protocol" viewpoint
>          - a valid scenario has been described that would 
> make use of (ii).
> Excluding it would be a useless restrictions. Would it not?
> Can you agree to the proposal of updating REQTS to allow 
> support of (ii).
> 
> 
> >and (iii) are required.
> 
> Your breakdown misses part of the discussion we've had which 
> is precisely 
> while I splitted that case earlier. Bandwidth is just one of 
> the Traffic 
> Trunk attributes. Based on the thread I think it is useful to 
> split (iii) into:
>          (iiia) Using E-LSPs with traffic from multiple OAs 
> with multiple 
> BW and single value for all other attributes (preemption, CT, 
> affinity...)
>          (iiib) Using E-LSPs with traffic from multiple OAs 
> with multiple 
> BW and multiple values for all other attributes (preemption, CT,...)
> 
> Those two cases do require different protocol extensions (eg signal 
> multiple preemtion/CT/affinity).
> 
> I thought we may be able to agree on (iiib). For example , I 
> understand 
> that while Sharham felt (iiia) should be added he agreed 
> (iiib) should not.
> So again, leaving aside (iiia) for now, can you agree that 
> (iiib) shoudl 
> not specifically be added to REQTS draft for now until SPs 
> have expressed 
> the requirement for it?
> 
> Cheers
> 
> Francois
> 
> >Kind Regards
> >Nabil
> 
>