[ntpwg] NTPv4 Draft issue [RATE]
rich.osman at nsn.com
rich.osman at nsn.com
Wed Jul 25 20:26:39 UTC 2007
Y'all:
Reference: RATE On Page 23 section 7.4 (b) says:
For kiss code: "RATE the client MUST immediately reduce its polling
interval to that server and continue to reduce it each time it receives
a RATE kiss code."
This seems to require a behavior that doesn't make sense to me. If I
parse this using definition 1) below it says that the client will poll
the root **more frequently** when I suspect the intent to get it to slow
down. The mailing list discussions I found seem to back this up.
I would change the text to replace the word 'interval' with the word
'rate.'
Webster's says:
Interval:
1 a : a space of time between events or states b British :
INTERMISSION
2 a : a space between objects, units, points, or states b :
difference in pitch between tones
3 : a set of real numbers between two numbers either including
or excluding one or both of them
4 : one of a series of fast-paced exercises interspersed with
slower ones or brief rests for training (as of an athlete)
And a point of curiosity: why isn't the magnitude of back off specified?
Looking at the mailing list traffic this was mentioned, but does not
seem to have been followed up. This seems to be a congestion control
mechanism. I can see someone backing off 1 ns per RATE packet and
claiming compliance. That gives them an advantage in a congested
environment. I suppose the server can DENY a client it doesn't see
backing off fast enough, but I can see lots of room for disagreement
depending on the focus of the developer. I'd use a factor of 10 or at
least a factor of two back off.
--
Rich Osman
Nokia Siemens Networks
Cell: +1 469-831-5513
SMS (email to phone) mailto:4698315513 at tmomail.net
GSM: +358 40 546 7374 (When out of US)
mailto:rich.osman at nsn.com
6000 Connection Drive; Irving, TX 75039
More information about the ntpwg
mailing list