One question would be how widespread is the practice of
disabling source routing, for machines that are outside of
firewalls. That would seem like the first experiment to do.
tom
--------
Date: Wed, 29 Jul 1998 16:33:48 -0700 (PDT)
From: Kenichi Ishikawa <ishi@cs.washington.edu>
To: tom@cs
Subject: Internet measurement (my interest)
Message-ID: <Pine.BSF.3.96.980729163100.22814A-100000@fsexp.cs.washington.edu>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Status: RO
Finally, I become to have an interest in the Internet measurement
area.
I am thinking to do the measurements described below.
Are they meaningful measurements?
If there is any other high prioritized thing should be done, let me
know.
[measure path characteristics between any two points in the Internet]
metrics:
RTT, drop rate, available band width and stability of them.
method:
The basic idea is the combination of 'bing' command and source
routing. 'bing' can measure path characteristics between any two
hosts ,HostA and HostB, only if HostA is on the routing path to
HostB.
If we alter the 'bing' to use source routing, any hostA can be
used as a nearest host, because we can establish routing path to
HostB via HostA by source routing.
Next figure depict new bing's behavior.
UW -------> HostA --------> HostB
Source path to
Routing measure
ICMP_ECHO #1 --------->
ICMP_ECHO #2 ------------------------->
= = = = = = = = = = = = = = = = = = = = = = =
and Guess this ---------->
Since bing calculates the difference between two ICMP_ECHO
packets, the measured result is less accurate than traceroute.
I'd like to compare the result of bing and the result of
traceroute server at first.
-- Kenichi Ishikawa