[trill] 答复: review of Trill over UDP/IP

Hi, Donald,

Thank you for the prompt reply. Please see inline.

-----邮件原件-----
发件人: Donald Eastlake [mailto:d3e3e3@gmail.com] 
发送时间: 2014年7月15日 15:52
收件人: Zongning
抄送: trill@ietf.org
主题: Re: [trill] review of Trill over UDP/IP

Hi,

On Mon, Jul 14, 2014 at 8:40 PM, Zongning <zongning@huawei.com> wrote:
> Hi, folks,
>
> (Re-send after subscription, sorry for potentially duplicated 
> message…)
>
> I’ve briefly reviewed the draft mainly from transport consideration’s 
> perspective, and have a question on congestion control.
>
> It says “When a TRILL over IP flow carries primarily IP-based traffic, 
> the aggregate traffic is assumed to be TCP friendly… and no additional 
> congestion avoidance action is necessary”. I’d like to see more 
> discussion about whether the adoption of UDP encapsulation by RBridges 
> between the end-hosts will negate such conclusion? Another word, even 
> if the end-hosts use TCP for congestion control, should there be some 
> congestion consideration between the RBridges that talk UDP?

Congestion control has been a very active area of discussion for other drafts. The congestion control text in the current version of draft-ietf-trill-over-ip was just copied with minor edits from an earlier version of draft-ietf-mpls-in-udp. That mpls draft has since been updated with substantial changes to its congestion section so I'm sure some work will be needed here.

[Ning] Then I'd suggest to add reference to [RFC5405] "Unicast UDP Usage Guidelines for Application Designers", which provides several cases in 3.1.3 about whether the UDP tunnel user should consider additional congestion control mechanism or not. Moreover, similar to draft-ietf-mpls-in-udp, I think it would be beneficial to add some brief text to clarify which case(s) in [RFC5405] that the draft-ietf-trill-over-ip belongs to, based on the specific features (if any) of Trill traffic.

> Plus, I’d know the reason why choosing UDP? I cannot find any in the 
> draft – but I am sure there are reasons discussed in the group. :-)

This was discussed at IETF 89 and the consensus there was confirmed on the mailing list. See http://www.ietf.org/mail-archive/web/trill/current/msg06206.html
You could look at the minutes at
http://www.ietf.org/proceedings/89/minutes/minutes-89-trill or listen to the audio recording.

Thanks,
Donald
=============================
 Donald E. Eastlake 3rd   +1-508-333-2270 (cell)
 155 Beaver Street, Milford, MA 01757 USA  d3e3e3@gmail.com

> Sorry if I misunderstood any point, and thanks for your clarification 
> in advance!
>
> -Ning
>
>
>
>
> _______________________________________________
> trill mailing list
> trill@ietf.org
> https://www.ietf.org/mailman/listinfo/trill
>