Hi Mikael,

> 
> Yes, I noticed. I still think that saying "X MUST be 4 (but 
> it may also be 
> 3)" is a weird thing to say.

Would it work if we rephrase the text in section 4 as follows:

Old:

Note that the PTP LSPs MUST only carry PTP messages and MAY carry MPLS/MPLS-TP 
control and management messages such as BFD and LSP-Ping.

New:

We should be selective about the kind of traffic that flows over PTP LSPs. All 
PTP messages MUST only be sent on the PTP LSP. It MAY, in addition to this also 
carry MPLS/MPLS-TP control and management messages such as BFD and LSP-Ping.

> 
> >> In section 10, I think it should be treated as control 
> plane traffic 
> >> with higher priority than what mostly is used for EF? Or 
> is EF (which 
> >> might carry a lot of voice traffic as well) enough?
> >
> > EF is better than the default. I don't think we can mandate 
> AF since 
> > that would require operators to provide assurance of 
> delivery as long as 
> > the traffic is not exceeding certain rate. Now, how does 
> define that 
> > rate?
> 
> I think you misunderstood.
> 
> In my world, control plane traffic (originated by the router 
> itself) may 
> have higher outgoing QoS priority than EF. AF is definitely 

Yes, its probably NC.

> further down 
> the list of priorities, both when it comes to jitter and packet loss 
> handling.
> 
> On Cisco for instance, people usually use the MQC "priority" 
> command for 
> EF, which means it's a low latency queue which is always 
> emptied first, 
> then there might be some round robin handling of other 
> queues. Is this 
> queue enough, or is it better to just handle it with low drop 
> probability 
> and just make sure the time information in the packet is 
> accurate and the 
> delay of the packet is of little importance?

I am still not sure if I understood your point. Are you suggesting that all 
such packets be marked as NC? 

Cheers, Manav

> 
> -- 
> Mikael Abrahamsson    email: [email protected]
> 
_______________________________________________
TICTOC mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/tictoc

Reply via email to