Sender: Andras.Veres@eth.ericsson.se
Message-ID: <3AD4895F.37615846@eth.ericsson.se>
Date: Wed, 11 Apr 2001 18:42:07 +0200
From: Andras Veres <Andras.Veres@eth.ericsson.se>
To: Sally Floyd <floyd@aciri.org>
CC: end2end-interest@postel.org
Subject: Re: [e2e] two questions about the Internet
References: <200104101715.f3AHFts03402@elk.aciri.org>


We have collected two traces on a small ISP 
network (33k modems). The original idea was
to measure the time delay between the first
TCP SYN and the ACK. Although this delay
is not strictly the RTT of the path, since
it contains possible SYN retransmissions and
the response delay of the server itself,
it is a good estimation of the RTT seen
from a modem user's perspective.

Sally: The logs can be used to investigate
delay statistics, and the correlation between
delay and the time of the day.
To correlate the delays with the end-to-end
path, e.g., get hop-counts, we would need
both IP addresses of the connection. I have 
all the necessary data to add the addresses
to logs, but first I will need authorization 
to release this info.
Anyway, for now, I can release the just the 
raw delays, see below.

The logs contain the raw data: in a 
<timestamp delay> format.

You can download the logs from:
http://www.comet.columbia.edu/~veres/DATASETS/Datasets.html


Andras