Re:Which clock to trust?

Jim Varney (jamesv@softcom.net)
Tue, 15 Sep 1998 22:24:10 -0700

On 15 Sep 98, Robert G Fenske Jr wrote:

> On Mon, 14 Sep 1998 dmbrierley@taz.dra.hmg.gb wrote:
> > I've been following this thread, and I fail to see how anyone can get
> > an accurate time over the Web, with uncertain delays in transmission 
> > and nodes.  Either a radio signal or the 1-900 number is preferable.
> 
> 	Is is quite possible to get times with 10s of milliseconds.  The
> NTP (Network Time Protocol) and associated programs have a lot of research
> and developement behind them to handle this kind of time transfer and


I agree, and have found Internet time service to be very good.  Robert's 
right, you need to separate the NTP time services from the little clocks 
you see on some web pages.

NTP has a distinct advantage over telephone time service, and that is that 
the NTP protocol actually measures network time delay and adjusts the time 
accordingly.

My (simplified) understanding is that the first data packet measures the 
round trip network delay.  The next data packet contains a time stamp and 
the measured delay to provide the predicted time at your computer.  The 
time error comes from variations in network delay and not from the delay 
itself.

For example, if I use the "ping" utility to this network time service:

clock.llnl.gov (128.115.14.97) 
Location: Lawrence Livermore National Laboratory, Livermore, CA 
Synchronization: NTP V3 primary (WWVB clock), Sun 3/60 

I get

  Pinging clock.llnl.gov [128.115.14.97] with 32 bytes of data:

  Reply from 128.115.14.97: bytes=32 time=297ms TTL=241
  Reply from 128.115.14.97: bytes=32 time=283ms TTL=241
  Reply from 128.115.14.97: bytes=32 time=285ms TTL=241
  Reply from 128.115.14.97: bytes=32 time=299ms TTL=241

  Ping statistics for 128.115.14.97:
    Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
    Approximate round trip times in milli-seconds:
    Minimum = 283ms, Maximum =  299ms, Average =  291ms

After adjusting for the 0.291 second network delay, you can see that the 
network deviation from packet to packet is on the order of 10 mS or so.

Your mileage will vary.

If you want to experiment: Dimension 4 is freeware at

http://www.accessone.com/~thinkman/dimension4/

 -- Jim

__________________________________________________________
Jim Varney
Member, Sacramento Valley Astronomical Society
Sacramento Iridiums www.softcom.net/users/jamesv/index.htm