Hi Adrian,
Thank you for your and Dhruv's suggestions which are very helpful for us. I
may not have given this issue enough consideration before the meeting.
Preliminarily, I agree with that a new UDP port may not be necessary. We will
update this part in the next version and collect the comments from the QUIC
Working Group. Thank you for your comments!
Best Regards,
Tingting Han
China Mobile
Mobile: +86 13810236960
e-mail: hantingt...@chinamobile.com
发件人: Adrian Farrel
时间: 2024/11/07(星期四)05:49
收件人: hantingting;
抄送人: pce;
主题: Thinking about PCEP over QUICHi Tingting,
Thanks for your presentation in today's meeting.
I wanted to expand on my comment about the use of a new UDP port number
specifically for PCEP over QUIC.
I am not a QUIC expert, and I suspect Dhruv isn't either, but we chatted
quickly and our understanding is the same. That is, the source and
destination UDP ports are not tied to the application, but it is the QUIC
connection ID that can be used to determine the use to which the connection
is put, with ALPN being the way that that information I exchanged.
My suggestion was to look at
https://datatracker.ietf.org/doc/draft-ietf-netconf-over-quic/ I think this
would be a good model for your PCEP work. After all, the behavior and use
case for PCEP is not dissimilar to Netconf. The Netconf draft does not
assign a UDP port number. I think that the Netconf working group discussed
this with QUIC experts to make sure they were doing the right thing.
However, I think you have modelled your approach on
https://datatracker.ietf.org/doc/draft-retana-idr-bgp-quic/ That draft does
suggest using a new port number. So I went to have a chat with Alvaro about
what is going on there. His explanation was that Dhruv and I have correctly
understood QUIC, but that BGP is a special case because it runs on the
critical path (both in the router and in the routing infrastructure).
Therefore, he claimed, it was necessary to use a different port number so
that BGP-over-QUIC could be easily distinguished from any other use of QUIC
allowing the router to prioritise BGP or block other applications. My claim
that this could easily be handled with the connection ID was answered by
Alvaro saying that the use of the port number allowed other uses to be
filtered out with less processing - this is possibly true, but I find it
debatable that there is much saving.
Anyway, I think that PCEP is closer to Netconf than to BGP (not BGP-LS).
A possible way forward would be to reach out to the QUIC working group and
ask their advice.
Cheers,
Adrian
_______________________________________________
Pce mailing list -- pce@ietf.org
To unsubscribe send an email to pce-le...@ietf.org
