[ntp:questions] Re: time

Terje Mathisen terje.mathisen at hda.hydro.com
Mon Nov 28 17:27:41 UTC 2005


K wrote:

> Hi,
> 
> 
> 
> What are some good ways to get time?
> 
> I know about NTP v3 and v4 and use it on some systems.  I need a time
> accuracy of < 1 ms.
> 
> 
> 
> NTP over gigabit Ethernet is down to what time of precision?   1ms?   Does
> anyone have any data on this?

I have been testing this for a few days here:

I have a machine (running SuSE linux 10.0) with a week-old version of
the ntp-dev code.

It is synced to two GPS-driven servers, both running FreeBSD 6.0, plus a
series of others servers, but they don't matter here.

One of the two source servers are on the same 100 Mbit LAN switch, while
the other server is located on a separate, fiber-connected & routed,
network.

The Linux machine is configured with both minpoll & maxpoll = 4, i.e. it
polls the reference servers every 16 seconds, generating about 100
bits/second of background traffic.

According to ntpq and the loopstats, it seems like the offset to the
LAN-attached server is in the low tens of microseconds, while the routed
server reports offsets 2-3 times larger, but still well below 100 us.

With lightly loaded Gig-E you should definitely see rms performance in
the sub-50 us range.

Terje

-- 
- <Terje.Mathisen at hda.hydro.com>
"almost all programming can be viewed as an exercise in caching"




More information about the questions mailing list