Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!UDEL.EDU!Mills
From: Mills@UDEL.EDU
Newsgroups: comp.protocols.tcp-ip
Subject: Re:  trace route to OZ
Message-ID: <8907241343.aa23745@huey.udel.edu>
Date: 24 Jul 89 17:43:53 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The Internet
Lines: 25

Sergio,

Some clue to the interesting behavior of our burgeoning Euroswamps may
lie in the far different behavior of the JvNC-Oslo circuit compared to the
JvNC-London circuit. The data I have here is collected with a Network
Time Protocol peer path between time servers in Delaware and Oslo and
over the last few weeks. The delays are about 850 ms and clock offsets
about 10 ms, not bad at all for any circuit, domestic or otherwise. The
servers ping each other about once every 17 minutes and run continuously,
so make a useful record of connectivity and congestion. If the path
becomes flaky the delay, offset or dispersion quickly reflect the fact,
so these time-server gizmos make good tripwires for detecting waves in the
swamps.

Okay, now the shoe drops. Jon, who would you like to bring up NTP on one
of your Gower Street munchkins? While at it, you might interest Robert
Cole at HP Labs in joining chimes. This way we could all watch each other's
clocks and get a better handle on (a) what is going on and (b) assess how
well precision time capability can help in cases like this. Come to think of
it, I'm sure you and Robert know each other and may have already discussed
this.

Perhaps further discussion should be offline.

Dave