Index Journeys
RFC 28 UNKNOWN ARPANET · TIME

Time Standards

A one-paragraph request from SRI: how accurate should a network-wide clock be?

On rfc-editor.org ↗
PUBLISHED
January 1970
AUTHORS
W.K. English
STREAM
Legacy
DOI
10.17487/RFC0028

RFC text

Original text · RFC Editor View on rfc-editor.org ↗ 
Network Working Group                                   Bill English
RFC-28                                                  SRI
                                                        13 January 1970




                          TIME STANDARDS




We are about to install a relatively accurate real-time clock on our
system.  I would like any comments relative to Network time standards
for such things as delay measurement so we can decide how good our clock
should be.














DISTRIBUTION:

A. Bhushan, MIT
S. Carr, Utah
G. Cole, SDS
S. Crocker, UCLA
K. Fry, MITRE
J. Heafner, RAND
B. Kahn, BB&N
T. O'Sullivan, Raytheon
L. Roberts, ARPA
P. Rovner, LL
R. Stoughton, UCSB

AI ELI5

On this page
PROBLEMSRI was about to install a relatively accurate real-time clock and did not know how good it needed to be for network jobs such as measuring delay.
MECHANISMThe RFC does not specify anything. It asks the other network sites for comments on Network time standards.
THE CATCHThe text records only the question and its distribution list. It contains no answer, so it sets no standard.

§1A Request for Comments

RFC 28, dated 13 January 1970 and written by Bill English of SRI, is a single short paragraph. SRI was about to install a relatively accurate real-time clock on its system.

Before choosing, the author wanted to hear from the rest of the network group about Network time standards, so the sites could decide together how good the clock should be.

Someone is about to buy a clock and asks his friends how precise it should be, so everyone can agree.

§2What the Clock Is For

The only use the RFC names is delay measurement. Measuring how long something takes to cross the network needs a clock, and the quality of the clock limits how good the measurement can be.

The RFC gives no figures, no required accuracy and no proposed method. It only says the answer to 'how good should our clock be' depends on what the network community wants from time standards.

If you want to time a race, your stopwatch has to be good enough to tell who won. The note asks how good the stopwatch must be.

§3Distribution and Outcome

The memo is addressed to the people who then made up the network group, and lists them with their sites: MIT, Utah, SDS, UCLA, MITRE, RAND, BB&N, Raytheon, ARPA, LL and UCSB.

There is no technical content beyond the request, so the RFC itself defines no time standard. Any replies are not part of this document.

The note was handed to a list of friends at different places. Their answers are not in the note.