[ntp:questions] Query about NTP accuracy

Unruh unruh-spam at physics.ubc.ca
Fri May 22 23:47:35 UTC 2009


Andy Yates <andyy1234 at gmail.com> writes:

>Unruh wrote:
>> Andy Yates <andyy1234 at gmail.com> writes:
>> 
>>> Does anybody have any figures that shows the effect on accuracy of an
>>> NTP v3 client using a stratum 1 server rather than a stratum 2 or 3
>>> server? It's all in a GE LAN based scenario, commercial stratum 1
>>> servers connected to GPS and stratum 2 and 3 servers are typically
>>> dedicated Linux boxes.
>> 
>>> The reasons is that I would rather scale by adding strata - its a very
>>> big data center with thousands of clients and has several "zones" that
>>> are isolated. However some opinion is suggesting we run IRIG-B between
>>> the GPS receiver and a bunch of stratum 1 servers and clients access
>>> these directly. Much more expensive and any increase in accuracy from a
>>> client experience may be negligible.
>> 
>>> However I'm been pressed to supply an SLA for accuracy. My argument is
>>> that although you can get your stratum one server to synchronize to
>>> microseconds of UTP, as soon as the client uses NTP v3 over the LAN,
>>> even a GE LAN, then the accuracy degrades and putting well designed well
>>> specified stratum between the boxes is not going to decrease accuracy
>>> sufficiently to warrant purchasing many stratum one appliances.
>> 
>> I think you need some specifications. What accuracy do you really need?
>> Without that you are simply wasting your, and everyones time and money.
>>  Direct GPS connection can give you
>> 1 usec. Lan connection to those can give you 20usec average. Then each
>> stratum will loose about 20-50 usec (my experience). Do you need better than msec

>Hi Unruh

>Totally agree about specification - trouble is the customers include
>third parties who don't issue specifications so we've been tasked to
>make it v.accurate, industry leading etc. however I'm not going to
>encourage blowing cash on diminishing returns. Thanks for the experience
>figures, this is exactly what I need.


Note also that it can depend on the network harware. When we ran 100Mb/s
switches, the delay time was astonishingly regular and repeatable. Now
that we have switched to Gb  switches it is not. It much worse. There
seems to be a one way "quantization" It used to be that the round trip
time from the machines to the ntp stratum 1 source was consistantly
140-160usec. Now it varies, with one level still at the 140usec, but
other round trip times of 280 or even longer (on the same machine).
 And they seem to be one
way delays ( the worst kind). Ie, you really need to select the switches
between your stratum 1 sources and your machines carefully to get the
best behaviour. 





More information about the questions mailing list