[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: [PWE3] Generic vs Preferred CW
Yaakov,
See below.
Thanks,
Matthew
"Yaakov Stein"
<yaakov_s at rad.co To: Mustapha.Aissaoui at alcatel.com, "pwe3" <pwe3 at ietf.org>, "Stewart Bryant"
m> <stbryant at cisco.com>
Sent by: cc:
pwe3-bounces at iet Subject: RE: [PWE3] Generic vs Preferred CW
f.org
06/05/2005 12:45
Mustapha,
> Yakov,
> You are right that the CW BCP is not just about the ECMP
> issue. I obviously have not re-read the introduction for some time :-)
OK, we agree on the first point.
> I however do not agree with the idea of a single CW for all
> applications. I know you have already proposed this, but in
> practice, each technology has its own specifics.
I can't agree more.
That's why the generic CW has te flag fields left for the "specifics".
MB> I think you mean the preferred CW.
To date there have only been 2 encap-dependent requirements
for the CW, namely flags (notably the defect flags in the TDM drafts
and the FECN/BECN flags in the FR draft) and pointers
(in the CEP draft).
Can you think of anything else that may be required
that can not be supplied by the generic CW?
MB> The context / semantics of the flags field is completely dependent on
the type of the PW. This is true for the length field as well. All this CW
does is provide a template. This is not a complete specification of the CW,
and cannot be if we insist on putting client information in the CW.
> I guess we
> (or some other people) will have to make a decision for each
> new application when it comes up.
But now we come to my second point, namely standardization.
Sure, maybe someone prefers some rearrangement of the fields,
or some other bit allocation. But the reason we need the IETF
and PWE is to reach agreement / compromise on these issues.
MB> And I believe that option (a) is a reasonable compromise that
recognises the fact that different current and future PW types may have
different requirements on the CW, while providing sufficient guidance that
future types should follow a common template.
Would you want the IP header to have the precise position of
the source address dependent on what the packet is carrying?
Why shouldn't the content of an MPLS packet determine
how many bits the label has?
MB> How is that relevant? the interpretation of the values in the CW is
dependent on the PW type.
MB>
Y(J)S
_______________________________________________
pwe3 mailing list
pwe3 at ietf.org
https://www1.ietf.org/mailman/listinfo/pwe3
_______________________________________________
pwe3 mailing list
pwe3 at ietf.org
https://www1.ietf.org/mailman/listinfo/pwe3