Re: [PWOT] RE: draft-danenberg-sonet-ces-mpls-mib-00.txt

Heiles,

>Concerning CEM over MPLS for CBR signals like a STS-SPE/VC in general I 
>share some of the doubts expressed in a previous mail, specifically for 
>clock jitter and wander. The MPLS transport could result in excessive 
>pointer justification of the STS-SPE or VT outside the Sonet/SDH 
>specifications.

As you have probably noticed, one of the areas that still needs more 
fleshing out in draft-malis-sonet-ces-mpls-03.txt are the underlying QoS 
requirements for the MPLS transport.  Of course, the QoS you can get is 
dependent on the capabilities of the MPLS LSRs - best effort forwarding 
only is obviously insufficient.  The more interesting question is what IS 
sufficient in the transport network to be able to meet the generally 
accepted requirements for bit error rates, clock jitter and wander, etc., 
and how large do the jitter buffers need to be at the receiver?

Of course, using the EF codepoint in either an E-LSP or L-LSP is the 
obvious starting point, but that will still be dependent on the network 
operators allocating sufficient resources for EF traffic in their networks 
so that the emulated SONET circuits perform adequately.  It may end up 
being an iterative process, at least at first, in order to get it right.

Regarding jitter buffers, most people I've discussed this with seem to 
think that about 10-15 ms worth of buffering should be more than sufficient.

I appreciate any other thoughts you may have.

Cheers,
Andy

_______________________________________________
pwot mailing list
pwot@ietf.org
http://www.ietf.org/mailman/listinfo/pwot