Um, the U in UDP stands for User. As in, userspace.

A kernel implementation of this hack? It would need users first. 

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Eddie 
Kohler
Sent: 10 January 2011 15:12
To: Pasi Sarolahti
Cc: [email protected]; 'dccp' working group; Colin Perkins
Subject: Re: [dccp] draft-ietf-dccp-udpencap-03 - 6-tuple

I am relatively neutral on this.  A 4-tuple is cleaner.  I worry that a 6-tuple 
would be a pretty major disincentive for kernel implementations of 
DCCP-over-UDP.  (Perhaps Gerrit has some feedback.)  I slightly prefer a 
4-tuple for this reason.

Eddie


On 1/10/11 6:36 AM, Pasi Sarolahti wrote:
> On Jan 3, 2011, at 6:34 PM, Colin Perkins wrote:
>
>>> ~ Following the WGLC there was debate on the 4/6-Tuple and how this would 
>>> work with different UDP and DCCP port values. As I see it, the current 
>>> proposal is to eliminate the 6-Tuple text and use only the outer UDP ports 
>>> for demultiplexing.
>>
>> If I understand what's proposed correctly, I don't think this would be a 
>> good idea. The ability to have a well-known UDP port on which tunnelled DCCP 
>> connections can be accepted seems important to me; as does the ability to 
>> run a server accessible via UDP and native DCCP listening on the same port, 
>> also accessible via tunnelled DCCP. Neither of these are possible if we use 
>> only the outer UDP ports.
>
> Fair point.
>
> The latest discussion started to wander off from discussing the 
> benefits/disadvantages of using the UDP/IP 4-tuple instead of 6-tuple, 
> but I don't recall anyone really opposing the 6-tuple text. Is someone 
> strictly against the 6-tuple model, or do we have rough consensus of 
> sticking with the current text in the draft? (with editorial changes 
> as suggested in earlier mails)
>
> - Pasi
>

Reply via email to