mailing list archives
Re: Non-GPS derived timing sources (was Re: NTp sources that work in a datacenter)
From: Jerry Scharf <scharf () vix com>
Date: Sat, 28 Jun 2003 08:36:43 -0700
--On 06/01/2003 11:57:21 PM -0400 Sean Donelan wrote:
On Sun, 1 Jun 2003, Marshall Eubanks wrote:
Every major time service and most national standards labs maintain a
set of clocks of comparable accuracy - US, UK, France, Germany, Russia,
Japan, Australia, etc., so there is no shortage of timing info to
compare it with.
Actually my question wasn't so much about other national standards labs,
but that almost every major Internet backbone worldwide seems to trace
their time source to GPS. Maybe not that surprising for US/North American
providers, but even non-american backbones seem to use GPS.
To be clear, I'm not talking about individuals syncing things to lots of
different clocks. Clock.ORG has lots of clock sources around the world.
I'm talking about what network providers use.
It was just one of those midnight projects a month or so ago, when I
noticed my carefully balanced selection of tickers had slowly over the
last few years all changed from other time sources to GPS. Probably
not critical, but national standards labs have accidentily flipped
the wrong switch in the past and done strange things to their time
broadcasts. Yes, lots of people noticed, and it was fixed quickly. NTP
has all this great logic for sanity checking time sources, but if they
all come from the same origin, what happens?
digging back through some old mail.
This is one of those like asking where we would get water after the
reservoir dam broke and flooded the town. The planes on GPS instrument
approach are going to be much less happy than you are. How about those
satellites that use GPS for attitude awareness. Or the rail anticollision
systems that are now GPS based. If GPS time gets screwed, internet time is
the least of our worries.
Some of us remember the false ticker that caused the introduction of the
NTP protection code. I don't think that anyone will ever build code like
the fuzzball that stopped routing when the time got confused (although I
guess I really shouldn't say that.)
As for GPS jamming, it's easy to do on a local basis but hard to do on a
widespread basis (unless you own the satellites.) For the accuracies that
most people care about, dropping one stratum won't be the end of the world.
Not too many people care about the microseconds (except us time geeks.)
As for why everyone is switching to GPS time: it's cheap, it's much more
accurate that anything else and it's available everywhere you can see a
reasonable amount of the sky with no service charge. Other than that I
can't see why people would do it.