[ntp:questions] strange behaviour of ntp peerstats entries.

Richard B. Gilbert rgilbert88 at comcast.net
Wed Jan 30 19:41:37 UTC 2008


Brian Utterback wrote:
> Richard B. Gilbert wrote:
> 
>> Maarten Wiltink wrote:
>>
>>> "Richard B. Gilbert" <rgilbert88 at comcast.net> wrote in message
>>> news:479F8551.5000505 at comcast.net...
>>> [...]
>>>
>>>> I can imagine an RTT of 60-70ms.  What I have difficulty imagining is
>>>> using such a source to synch with.
>>>
>>>
>>>
>>> Pfft. Kids these days.
>>>
>>> (There's nothing wrong with an RTT of 60 to 70 ms per se. For 
>>> example, if
>>> it were always exactly 65 ms, that would probably be an *excellent* time
>>> source. The problem is the jitter, and as the example of the four 
>>> possible
>>> paths along the two possible routes shows, even that can, under the 
>>> right
>>> circumstances, be solved.)
>>>
>>> Groetjes,
>>> Maarten Wiltink
>>>
>>>
>>
>> Well, since the maximum error in transmitting time from server to 
>> client is one half the round trip delay, it is usually wise to try to 
>> minimize that delay.  A server in Tokyo might have the time correct to 
>> within 50 nanoseconds but that does me little good in New Jersey!  The 
>> network path would be so long and pass through so many routers and 
>> switches that by the time it gets to me, the uncertainty will be a 
>> substantial fraction of a second.
>>
>> Usually, the number of possible paths will be far greater than four!
>>
>> The ultimate test is the actual performance under the stated conditions.
>> Based on general principles, the stated conditions are NOT where I 
>> would look first for best performance.
>>
> 
> Which is why NTP prefers the source with the smallest delay. The system
> I am using has servers whose delays are 51ms to 94. I can't find any
> closer. On my company LAN, the delays range from 16ms to 87ms. The
> offsets of all these servers agree to within 9ms. Sure, I am not
> going to get sub-millisecond from that, but I think it is probably
> more typical than your set-up.
> 
> Brian Utterback

I think NTP uses more than the delay to select a server.  ISTR that it's 
one half the round trip delay plus something else.  The resulting 
quantity is called the "synchronization distance".  I have forgotten 
just what the something else is and I'm too lazy to look it up.





More information about the questions mailing list