[geeks] ntp rant

[Greg A. Woods]

[ On Thursday, January 10, 2002 at 00:50:37 (-0500), Dave McGuire wrote: ]
> Subject: Re: [geeks] ntp rant
>
> On January 9, Greg A. Woods wrote:
> > I personally don't see why anyone would want better NTP than can already
> > be provided by syncing to any of the many available stratum-2 servers.
> 
>   Thank heaven *everyone* doesn't have that attitude, otherwise
> there'd be no stratum-1 servers.

Strictly speaking GPS alone does not a stratum-1 clock make.....  Sure
it's but one hop away from a real clock, in theory, but that's one
awfully big hop, even if you do have three satellites visibile in the
constellation at all times.  (Not to mention that as I understand it
there are still provisions for the GPS time base being scrambled a bit
more than usual when necessary.)

What's really stupid is that half those GPS-based ones claiming to be
stratum-1 are nowhere near so stable as to deserve the designation
(incl. our own GPS-based ntp.planix.com!)

>   Nah.  Good rubidium primary standards can be had for $1K or so, and
> rubidium oscillator modules can be gotten for about $400.

I've never seen one so cheaply that anyone serious about time standards
would ever settle for, but then most of the people I know who are so
serious are those providing time signals for television or telecom,
and/or are designing, building, or repairing the equipment to do so.  At
least not one that's been calibrated and certified!  ;-)

>   What's "normal"?  I suspect there are very few "normal" people here.

"normal" in this case is someone running a serious network of several
computers, and perhaps a small WAN as well.  Eg. me, maybe you, etc.

A television station, or maybe someone running a wide area carrier based
network, might need a highly stable clock to stay running even when
everything else goes to pot so that they can sync up with their
neighbours again when things come together.

If you or I are not on the net then the stability of our clocks at the
time don't matter a hoot.  If you are on the net you can have your
clocks at least as closely sync'ed to the real stratum-1 clocks as mine
are without anything more than a decent server and a copy of ntpd.

> Many of us hack on all sorts of different things that you may or may
> not be interested in.  I can assure you, however, that anyone who
> wants to do something like this probably has something better in mind
> than the output of the "date" program.

Exactly the point -- which is why I mentioned my colleague's rather
expensive and fruitless experiments with a GPS-based time source.

>   Personally, I built most of a cesium-based stratum-1 server because I
> *could*.  I had all the analog hardware for running calibrations of
> secondary frequency standards, so why not do a little extra work and
> make a *quality* stratum-1 ntp server as well?

Now that I can appreciate, but that's a whole long ways away from what
your average GPS-based stratum-1 server is, including ntp.planix.com.

-- 
								Greg A. Woods

+1 416 218-0098;  <gwoods at acm.org>;  <g.a.woods at ieee.org>;  <woods at robohack.ca>
Planix, Inc. <woods at planix.com>; VE3TCP; Secrets of the Weird <woods at weird.com>