[ntp:questions] .1 Microsecond Synchronization

Richard B. Gilbert rgilbert88 at comcast.net
Thu Jun 4 18:54:29 UTC 2009


ScottyG wrote:
> Hello.
> 
> The company I am working for needs to be able to record timestamps in a trading 
> system logs down to a .1 microsecond accuracy.
> 
> We will have servers located in London, New York and Chicago. There will be a 
> dedicated resilient link between London and New York.
> 
> Searches on the web have made claims that NTP can achieve this accuracy. 
> Unfortunately the sales rep for the NTP server we looked at told me that the best I 
> could expect is 2-5 ms synchronization across servers.

That sales rep was being realistic!!!  2-5 milliseconds is very good!
It's about what you can expect using NTP over the internet and querying 
nearby stratum 1 servers.

If you actually need 0.1 microsecond accuracy, be prepared to spend BIG 
BUCKS!!!!  The only means I can think of involve having your own cesium 
(atomic) clock. A rubidium clock might suffice!!  The accuracy and 
stability are poorer but it's much cheaper.

You load it into your truck or station wagon with generator and/or 
battery pack to sustain it, take it to the National Institute of 
Standards and Technology and have them set it to the correct time.  If 
you can keep it powered continuously, you will probably not need to have 
it calibrated more often than every five to ten years.

I think cesium clocks may start at the $50,000 level.  It might be more.
As a hobbyist, I'm quite certain that I can't afford it.

Then there is the problem of getting the time into your computer.  The 
latencies are difficult to measure!







More information about the questions mailing list