RE: [PWE3] PWE3 new WG items? PW QoS

["Busschbach, Peter B (Peter)" <busschbach at lucent.com>]

> -----Original Message-----
> From: Wei Luo [mailto:luo@cisco.com]
> Sent: Friday, February 20, 2004 6:20 PM
> To: Busschbach, Peter B (Peter)
> Cc: 'Proch, Daniel'; 'Rahul Aggarwal'; benjamin.niven-jenkins@bt.com;
> hshah@ciena.com; pwe3@ietf.org
> Subject: Re: [PWE3] PWE3 new WG items? PW QoS
> 
> 
> Hi Peter,
> 
> Actually we are in agreement on what specific issues need to 
> be resolved for PW 
> stitching.  We literally listed the same stuff.  What we 
> might disagree is how 
> to solve these problems.
> 
> You seem to be advocating that RSVP is the cure for all these 
> issues, 

To be precise: I am advocating that RSVP is *one* cure for all these problems, with the nice property that it exists today. Of course we can define another solution to PW-splicing from scratch, but I think it is preferable to use an existing protocol.

Otherwise, I agree with you that we need to define the scope of the problem before we can decide on a solution.


> but I am 
> more inclined to have the working group define the scope of 
> the problems first, 
> and then we could decide whether they can be solved by 
> implementation specific 
> mechanisms or a new protocol procedure is necessary.  The 
> current protocol 
> procedures are sufficient for multi-semgent PWs signaling in 
> the sense that they 
> can handle end-to-end setup, teardown, notification, and etc. 
> with proper 
> implementation at the stitching points.
> 
> ---Wei
> 
> Busschbach, Peter B (Peter) wrote:
> >>Actually PW stitching can be implemented entirely using 
> >>existing pseudowire 
> >>signaling procedures (using LDP or L2TP) without any new 
> >>protocol or extension. 
> >>  The signaling protocol is not an issue.
> > 
> > 
> > Hi Wei,
> > 
> > I respectfully disagree. The fact that we have a signaling 
> protocol to set up single PW segments doesn't mean that it is 
> clear how to establish multi-segment PWs. E.g.
> > 
> > - Are segments set up in a random order or is there a need 
> for a certain sequence of message?
> > 
> > - Is it necessary to use the same values for Interface 
> Parameters on each segment?
> > 
> > - Is VCCV used on single segments or end-to-end? If the 
> latter, do intermediate points monitor or do they blindly 
> forward the packets?
> > 
> > - In case a segment goes down, are alarm signals sent to 
> the end-points of the multi-segment PW or is the error 
> handled locally?
> > 
> > All these, and probably more, need to be specified. On the 
> other hand, if we decide that a multi-segment PW is 
> essentially an LSP that can be set up with RSVP-TE, the 
> RSVP-TE spec provides most of the answers.
> > 
> > Peter
> > 
> > 
> >>With regard to PW stitching, this working group probably 
> >>should focus on the 
> >>operational issues with regard to PW stitching, such as 
> >>whether Attachment 
> >>Circuit interworking is needed at the stitching point, how to 
> >>interwork between 
> >>different PSN types, sequencing, multi-hop VCCV, Inter-AS 
> >>considerations and 
> >>other network management subjects.
> >>
> >>---Wei
> >>
> > 
> 
> 

_______________________________________________
pwe3 mailing list
pwe3@ietf.org
https://www1.ietf.org/mailman/listinfo/pwe3