[ntp:questions] Confused about pool rotation
elliott.ch at verizon.net
Thu Feb 7 11:19:22 UTC 2013
Have you ever tried to synchronize to the pool servers with NTPD for any
length of time? I did once, admittedly several years ago. They weaved
and waved all over the time domain so much (at least +- 50 ms) that it
was impossible to have stable time on the client. I would suggest you
try synchronizing your home (or another) computer to the pool servers
with NTPD for a couple of days (and record with peerstats and/or
loopstats logs and inspect the results in a spreadsheet and graph)
before you design them into your app.
Using the pool servers are the "right" thing to do for your application.
However, you may be disappointed in the results. I use stratum 2 servers
with some success; stratum 1 servers are
way too overloaded, at least in the US.
The other side of the coin is that non-pool servers come and go with
distressing frequency. We lost several located on the East Coast of
the US around the time of Hurricane Sandy.
> -----Original Message-----
> From: questions-bounces+elliott.ch=verizon.net at lists.ntp.org
> [mailto:questions-bounces+elliott.ch=verizon.net at lists.ntp.org] On
> Behalf Of Robert Scott
> Sent: Wednesday, February 6, 2013 8:34 PM
> To: questions at lists.ntp.org
> Subject: Re: [ntp:questions] Confused about pool rotation
> On Wed, 06 Feb 2013 16:10:44 -0800, E-Mail Sent to this address will
> be added to the BlackLists <Null at BlackList.Anitech-Systems.invalid>
> >Considering that iPhones, and android phones can sync with
> > NTP servers just fine without an extra APP,
> > and the SDKs have monotonic nanosecond timers already?
> As I explained in another thread last week, my app is not a time app.
> It is a precision audio frequency measuring app. The users of this
> app expect frequency accuracy on the order of 12 ppm. To achieve this
> accuracy I cannot rely on the nominal 100 ppm accuracy of the quartz
> crystal that drives the device's audio sampling rate. So I need to
> perform an initial calibration to determine each device's actual audio
> sample rate, which is quite stable, and which is totally independent
> of the system clock. For the past two years my app has been doing
> this by having the user call the NIST telephone service and letting
> the device listen to standard 500 Hz or 600 Hz tones to determine the
> calibration factor. But this is a hassle. So I am trying to achieve
> 12 ppm frequency calibration by developing a temporary local time
> stamp that is disciplined by SNTP. I figure two SNTP samples about 2
> to 4 hours apart will give the required frequency accuracy, assuming
> typical SNTP jitter. This only needs to be done once since ongoing
> time accuracy is not my goal. And I hope to automate the process so
> the user can just start the calibration just before retiring for the
> night. The next day my app will be calibrated on his device and the
> user will not need to do that again.
> > I don't see how $300 can be considered low cost.
> It is quite simple. There are no comparable apps to mine that cost
> less. There are two comparable apps to mine that cost quite a bit
> > Not to mention phones with GPS, can get stratum 0 time
> > right from the satellite, without the asymmetrical
> > network latency.
> > Cell towers are already synchronized with stratum one
> > servers and / or GPS, and usually contain stable oscillators
> > e.g. rubidium, cesium, ...
> > Cell towers can't do seamless handoffs for moving cell phones
> > to adjacent towers without close synchronization.
> My app also has to run on iPads without cell tower access.
> questions mailing list
> questions at lists.ntp.org
More information about the questions