Hi Chris,

CCID 3 looks good to me, I'm OK with the text. 

We could just use the port number, 6514, as the service code. Since the
service identifier applies to more than DCCP, it probably makes more
sense the follow the scheme defined in RFC4340 where a 4 letter string
is used as the service identifier, such as the following:

SC:SYLG
SC=x53594C47
SC=1398361159

Cheers,

Joe
> -----Original Message-----
> From: [email protected] [mailto:[email protected]] On
Behalf
> Of Chris Lonvick (clonvick)
> Sent: Monday, April 12, 2010 6:40 AM
> To: [email protected]
> Subject: Re: [Syslog] AD review comments for draft-ietf-syslog-dtls
> 
> Hi Folks,
> 
> I'll suggest CCID 3 because that's my lucky number.  ;-)
> 
> Seriously, here is a relevant point from RFC 5238:
> ===vvv===
>     In addition to the retransmission issues, if the throughput needs
of
>     the actual application data differ from the needs of the DTLS
>     handshake, it is possible that the handshake transference could
leave
>     the DCCP congestion control in a state that is not immediately
>     suitable for the application data that will follow.  For example,
>     DCCP Congestion Control Identifier (CCID) 2 ([RFC4341]) congestion
>     control uses an Additive Increase Multiplicative Decrease (AIMD)
>     algorithm similar to TCP congestion control.  If it is used, then
it
>     is possible that transference of a large handshake could cause a
>     multiplicative decrease that would not have happened with the
>     application data.  The application might then be throttled while
>     waiting for additive increase to return throughput to acceptable
>     levels.
> 
>     Applications where this might be a problem should consider using
DCCP
>     CCID 3 ([RFC4342]).  CCID 3 implements TCP-Friendly Rate Control
>     (TFRC, [RFC3448])).  TFRC varies the allowed throughput more
slowly
>     than AIMD and might avoid the discontinuities possible with CCID
2.
> ===^^^===
> 
> My reasoning for choosing CCID 3 is that when some devices start up
they
> will queue up syslog messages until the network is up, and then they
will
> start to deliver them.  I don't want a large handshake to throttle
that
> initial burst of messages.  (Please challenge this assumption if you
have
> a better understanding of the process.)
> 
> I'll suggest that the specific wording will need to be: "MUST
implement
> CCID 3 and SHOULD implement CCID 2 to ensure interoperability".  Does
that
> sound OK to everyone?
> 
> 
> Joe: can you look at Sean's second question and let us know about
that?
> 
> Thanks,
> Chris
> 
> On Thu, 8 Apr 2010, Sean Turner wrote:
> 
> > I have one major comment and it relates to DCCP:
> >
> > The DCCP chairs tell me that to specify the use of DCCP the ID needs
to
> > decide which CCID it will use (CCID 2 is AIMD and CCID 3 is TFRC).
I
> was
> > hoping that the DTLS over DCCP RFC addressed this, but that RFC
doesn't
> pick
> > one it leaves this choice to the "application".
> >
> > Can you also confirm that the Port # is used as the DCCP service
code?
> >
> > spt
> >
> >
> >
> > _______________________________________________
> > Syslog mailing list
> > [email protected]
> > https://www.ietf.org/mailman/listinfo/syslog
> >
> _______________________________________________
> Syslog mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/syslog
_______________________________________________
Syslog mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/syslog

Reply via email to