Re: 500ppm - is it too small?
- From: mayer@xxxxxxx (Danny Mayer)
- Date: Thu, 20 Aug 2009 17:43:54 GMT
Harlan Stenn wrote:
Richard> I can't follow Dave's math but I'm reasonably sure that there is aIn article <TLSdnQ2E26bBLBnXnZ2dnUVZ_sydnZ2d@xxxxxxxxxxxx>, "Richard B. Gilbert" <rgilbert88@xxxxxxxxxxx> writes:
Richard> good reason for the 500 PPM limit. Since almost all computer
Richard> clocks can meet this criterion I'm not going to worry about it.
It's been a long time since I looked, but as I recall the 500ppm limit was
chosen to be more than sufficient to handle the vast majority of non-broken
hardware, and the value also plays a part in the "system constraints" and
the Allan intercept.
What I remember from the last time I looked at it was "changing the 500ppm
constraint will mean a Lot Of Work for lots of other places, and it's really
not worth the trouble."
If you read Dave Mills's book you will see that he discusses this in the
first chapter. The value of 500ppm was chosen arbitrarily but as you say
it handles the vast majority of cases. The point that he makes, however,
is that some limit needs to be chosen so that the algorithms used be
stable and converge. While you can change the limit, you then need to
review the all of the complex mathematics involved to ensure stability
and convergence since there will almost inevitably be some conditions
were some arbitrarily chosen limit will cause instability. See chaos
theory. In theory you can make the value a configuration option but the
results would not be guaranteed to do what you want.
If that is really not the case for Windows, then somebody who has a much
better "math brain" that I do should run all the numbers and propose an
alternative.
Windows is not really any different from any other operating system and
it has more to do with the system clock being used (that and interrupts)
than anything else. I regularly put my Windows laptop on standby, put
into hibernate and on and off mains power. None of this affects ntpd so
much that it cannot recover. The maths involved really has nothing to do
with it. Bill Unruh will no doubt point out that Chrony does a better
job of recovering from all this and that may to true, but ntpd is not
failing here.
Danny
--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
.
- References:
- Re: 500ppm - is it too small?
- From: Ulrich Windl
- Re: 500ppm - is it too small?
- From: David J Taylor
- Re: 500ppm - is it too small?
- From: Richard B. Gilbert
- Re: 500ppm - is it too small?
- From: David J Taylor
- Re: 500ppm - is it too small?
- From: nemo_outis
- Re: 500ppm - is it too small?
- From: Richard B. Gilbert
- Re: 500ppm - is it too small?
- From: Harlan Stenn
- Re: 500ppm - is it too small?
- Prev by Date: Re: So what? (Re: 500ppm - is it too small?)
- Next by Date: Re: So what? (Re: 500ppm - is it too small?)
- Previous by thread: Re: 500ppm - is it too small?
- Next by thread: Re: 500ppm - is it too small?
- Index(es):
Relevant Pages
|
