Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

GPS NTP Services

5 views
Skip to first unread message

ElfMagic

unread,
Mar 11, 2002, 9:15:22 PM3/11/02
to
Could someone please tell me if there is a GPS unit that will provide
NTP services and support equipment using Time Synchronisation using
the TCP/UDP Time Protocol (RFC 868), NTP v3 (RFC 1305), or Simple NTP
v4 (RFC 1769)?

We have a mixture of NT4 and Windows 2000 Servers, RS6000 IBM-AIX
servers, and Cisco routers/switches, and Lucent Cajun switches. All
these use different NTP protocols so I'm hoping to find a GPS product
that can support the above three NTP protocols.

At this moment we've been running a Windows NT server sync'ing with a
time clock out in the Internet using SNTP. But with us bringing
eCommerce business into a live system, I'd feel more comfortable about
having NTP internally and using the Internet's time clock as a
fail-over source.

Any help or opinions would be greatly appreciated.


Thank you,

Elvio Serrao

Koos van den Hout

unread,
Mar 12, 2002, 8:24:56 AM3/12/02
to
ElfMagic <elvio....@firstdata.com.au> wrote:
> Could someone please tell me if there is a GPS unit that will provide
> NTP services and support equipment using Time Synchronisation using
> the TCP/UDP Time Protocol (RFC 868), NTP v3 (RFC 1305), or Simple NTP
> v4 (RFC 1769)?

Yes,

http://www.meinberg.de/english/products/lantable.htm

(just a happy customer)

Koos

--
Koos van den Hout, PGP keyid 0x27513781
ko...@cs.uu.nl herding Suns and networks
+31-30-2534104 Visit my site about books with reviews
http://idefix.net/~koos/ http://www.virtualbookcase.com/

Marc Brett

unread,
Mar 12, 2002, 8:43:28 AM3/12/02
to
ElfMagic <elvio....@firstdata.com.au> wrote:
> Could someone please tell me if there is a GPS unit that will provide
> NTP services and support equipment using Time Synchronisation using
> the TCP/UDP Time Protocol (RFC 868), NTP v3 (RFC 1305), or Simple NTP
> v4 (RFC 1769)?

See: http://www.eecis.udel.edu/~ntp/hardware.html

--
Marc Brett +44 20 8560 3160 WesternGeco
Marc....@WesternGeco.com 455 London Road, Isleworth
ICBM: TQ 15774 76378 (OSGB) Middlesex TW7 5AA UK

Don Payette

unread,
Mar 12, 2002, 1:53:15 PM3/12/02
to
Check out http://www.truetime.com.

A happy customer.


elvio....@firstdata.com.au (ElfMagic) wrote:

-----------
Don Payette
Unisys Corporation
I speak only for myself; not my employer
Please reply in the newsgroup. Don't try
sending e-mail.

ElfMagic

unread,
Mar 12, 2002, 7:38:55 PM3/12/02
to
Don Payette <Nob...@nowhere.com> wrote in message news:<8kjs8uc2qbq6alqmn...@4ax.com>...

> Check out http://www.truetime.com.
>
> A happy customer.
>

Being predominately a Windows NT/2000 shop were you all able to use
the Windows 2000 Time Service (w32time) to sync with the NTP hardware
or did you require the rollout of a time sync service to all machines
that's proprietary to the NTP hardware?

Thank you,

Elvio Serrao

Don Payette

unread,
Mar 13, 2002, 12:59:34 PM3/13/02
to
Generally you don't want to use Microsoft's software as a
time server, it's ok as a client when talking to a hardware
time server, or a machine running the ntp daemon available
at http://www.ntp.org.

For example, we have the truetime box inside our firewall that
serves a number of machines running the ntp daemon (some Unix
some Windows). The corporation gets time from these servers.

For my workstation, I use Dimension 4 (http://www.thinkman.com).
For our mainframes, I wrote an ntp client that talks to any ntp server.
I've noticed, however, that when configured to talk to the
Microsoft implementation, the packets don't make it through
the validation I put into my client. So we recommend our
customers use the ntp daemon (which runs fine on Windows)
as their time server software.

elvio....@firstdata.com.au (ElfMagic) wrote:

-----------

David L. Mills

unread,
Mar 13, 2002, 4:05:45 PM3/13/02
to
Don,

At least in XP, Microsoft has broken rfc2030 (SNTP) and rfc1305 (NTP)
protocol specifications. XP clients send symmetric-mode packets instead
of expected client-mode packets. If one of these shows up on your server
doorstep and you have intentionally defeated cryptographic
authentication, this will fire up a symmetric-mode association and
potentially warp your server clock.

Compliant and authentication-enabled (default) NTP servers will in fact
(only) respond with server-mode packets on the Postellian principle to
be lenient in what to accept.

I have to conclude, since the violation is so blatent, that the intent
of Microsoft is to create a potential incompatability that could
preclude the use of non-Microsoft products.

Dave

Jonathan Buzzard

unread,
Mar 13, 2002, 5:36:06 PM3/13/02
to
In article <3C8FBF29...@udel.edu>,

"David L. Mills" <mi...@udel.edu> writes:
> Don,
>
> At least in XP, Microsoft has broken rfc2030 (SNTP) and rfc1305 (NTP)
> protocol specifications. XP clients send symmetric-mode packets instead
> of expected client-mode packets. If one of these shows up on your server
> doorstep and you have intentionally defeated cryptographic
> authentication, this will fire up a symmetric-mode association and
> potentially warp your server clock.
>
> Compliant and authentication-enabled (default) NTP servers will in fact
> (only) respond with server-mode packets on the Postellian principle to
> be lenient in what to accept.
>
> I have to conclude, since the violation is so blatent, that the intent
> of Microsoft is to create a potential incompatability that could
> preclude the use of non-Microsoft products.
>

Add in the fact that Windows NT based systems only keep time to the
nearest 10ms and you might as well forget any Microsoft OS as a
suitable platform for accurate timekeeping.

JAB.

--
Jonathan A. Buzzard Email: jona...@buzzard.org.uk
Northumberland, United Kingdom. Tel: +44(0)1661-832195

Don Payette

unread,
Mar 15, 2002, 11:36:26 AM3/15/02
to
I gracefully defer to Dr Mills. Don't use Microsoft time
software for anything. :-)

"David L. Mills" <mi...@udel.edu> wrote:

>Don,
>
>At least in XP, Microsoft has broken rfc2030 (SNTP) and rfc1305 (NTP)
>protocol specifications. XP clients send symmetric-mode packets instead
>of expected client-mode packets. If one of these shows up on your server
>doorstep and you have intentionally defeated cryptographic
>authentication, this will fire up a symmetric-mode association and
>potentially warp your server clock.
>
>Compliant and authentication-enabled (default) NTP servers will in fact
>(only) respond with server-mode packets on the Postellian principle to
>be lenient in what to accept.
>
>I have to conclude, since the violation is so blatent, that the intent
>of Microsoft is to create a potential incompatability that could
>preclude the use of non-Microsoft products.
>
>Dave
>

-----------

Don Payette

unread,
Mar 15, 2002, 11:43:23 AM3/15/02
to
Accuracy is in the eye of the beholder.

To much of the world (largely outside of this newsgroup), 10 ms
accuracy is amazingly good.

I was recently involved with a project to upgrade the accuracy
of one of the largest banking systems in the world. Their existing
system had time accuracy to seven seconds, and their requirement
was to improve that to one second.

Many people would be impressed if their PC just kept better time
than their watch.

Just providing a different perspective. :-)


jona...@buzzard.org.uk (Jonathan Buzzard) wrote:

>
>Add in the fact that Windows NT based systems only keep time to the
>nearest 10ms and you might as well forget any Microsoft OS as a
>suitable platform for accurate timekeeping.
>
>JAB.

-----------

Laurence F. Sheldon, Jr.

unread,
Mar 15, 2002, 12:21:42 PM3/15/02
to
On Fri, 15 Mar 2002, Don Payette wrote:

> Accuracy is in the eye of the beholder.
>
> To much of the world (largely outside of this newsgroup), 10 ms
> accuracy is amazingly good.
>
> I was recently involved with a project to upgrade the accuracy
> of one of the largest banking systems in the world. Their existing
> system had time accuracy to seven seconds, and their requirement
> was to improve that to one second.
>
> Many people would be impressed if their PC just kept better time
> than their watch.
>
> Just providing a different perspective. :-)

When I first got interested in NTP several years ago, my goal was to
keep a bunch of routers in agreement with the log server as to the
year, month and day.

Having them agree on hour, minute and second has been a boon.

Hal Murray

unread,
Mar 16, 2002, 6:10:52 AM3/16/02
to
>When I first got interested in NTP several years ago, my goal was to
>keep a bunch of routers in agreement with the log server as to the
>year, month and day.

I saw a really nice article several (many?) years ago. I think it was
in the HP Journal, but I can't find it at either HP or Agilent.

The story was roughly...

A power company asked their atomic-clock division about good clocks.
Suppose lightning takes out a big transmission line.
If they could record the time that circuit breakers
disconnect to within a few hundred nanoseconds and get the
data at both ends of the line, then they could compute where
along the line the problem was and send the repair trucks
directly to the right place rather than searching for the
problem.

HP produced a box containing a GPS unit and a good crystal.
(and a lot of software)

Anybody else see that article? Anybody got a URL that works?

--
These are my opinions, not necessarily my employer's. I hate spam.

Hans Jørgen Jakobsen

unread,
Mar 16, 2002, 11:22:04 AM3/16/02
to
In a pile of HP Journal from feb 1995 to aug 1997 i found in dec 1996
p 60: The Global Positioning System and HP SmartClock by John A Kusters.
It mentions with no details: "The HP59551A GPS measurements synchronization
module is designed to meet specific needs of the power generation an
distribution commuity."

A search on Google for "measurements synchronization power generation"
gave among other things:
http://www.symmetricom.com/technology/download/APP_TWFLPTS.pdf
/hjj

David L. Mills

unread,
Mar 16, 2002, 2:29:30 PM3/16/02
to
Hal,

A more common application is to use GPS to measure the actual 60-Hz
phase in different places in the network as a load monitor. I once
proposed a LORAN-C receiver for that purpose and actually built one for
test, but it wasn't accurate enough.

Dave

David L. Mills

unread,
Mar 16, 2002, 2:29:18 PM3/16/02
to Hal Murray
Hal,

A more common application is to use GPS to measure the actual 60-Hz
phase in different places in the network as a load monitor. I once
proposed a LORAN-C receiver for that purpose and actually built one for
test, but it wasn't accurate enough.

Dave

0 new messages