ietf
[Top] [All Lists]

RE: Date and Time on the Internet: Timestamps to Proposed Standar d

2001-11-14 07:40:03
My view is that accuracy requirements for timestamps are a matter for the
application.  E.g. in an email archive, a timestamp that is accurate within
a few minutes may be OK, for an e-commerce transaction log I would hope for
sub-second accuracy.  I don't think it is something this document can
usefully discuss.

The purpose of this memo is to provide an easily referenced, easily used
format for timestamps in Internet protocols that captures the kind of
requirements that occur over a range of common network applications.  It is
not really intended to be a treatise on the subtleties of time measurement,
nor to meet all of the requirements of specialized time applications
(though I would hope that a simple specification of this kind will make it
easier for the value of such specialized applications to be better
realized).  We also included some non-normative material that was felt to
be useful background, without (I hope) committing too much space to arcane
matters.

Section 4.4 was felt to be relevant because communicating hosts in
different timezones is an everyday reality of Internet applications.  Note
that the 'MUST' to which you refer is part of one of a number of approaches
that are suggested.

(By way of history:  Chris wrote an original document some time ago,
including a quantity of more specialized material, concerning some of which
it was proving difficult to achieve stable consensus.  Meanwhile, I had
found the timestamp elements of Chris' document were a very useful,
accessible reference for "everyday" protocol use, and suggested that we
pull these into a separate, and hopefully non-controversial, document that
could be published as an RFC and be available for wider use.)

#g
--

At 08:52 AM 11/13/01 -0800, Paul Skoog wrote:

Hello,

I have just reviewed Date and Time on the Internet:Timestamps
<draft-ietf-impp-datetime-05.txt>.  Are there any accuracy considerations
with respect to these timestamps?  Do they need to be accurate to some
level of precision (minutes, seconds?) or is the standard attempting to
define a format only?  Section 4.4 Unqualified Local Time asserts machines
using local time MUST synchronize with a UTC source.  NTP is asserted as a
mechanism to obtain UTC and it follows that this introduces some degree of
accuracy.  Should this be defined?  I am very interested in your thoughts
on this.

Thanks,

Paul Skoog
Product Manager, IP Network Timing
TrueTime, Inc.
3750 Westwind Blvd. Santa Rosa, CA  95403
t 707.636.1955 f 707.527.6640 c 707.481.1476
pskoog(_at_)truetime(_dot_)com / www.truetime.com

The information transmitted is intended only for the person or entity to
which it is addressed and may contain confidential and/or privileged
material.  Any review, retransmission, dissemination or other use of, or
taking of any action in reliance upon, this information by persons or
entities other than the intended recipient are prohibited.   If you
received this in error, please contact the sender and delete the material
from any computer.

------------
Graham Klyne
GK(_at_)NineByNine(_dot_)org



<Prev in Thread] Current Thread [Next in Thread>