[53950] in North American Network Operators' Group
RE: Network integrity and non-random removal of nodes
daemon@ATHENA.MIT.EDU (W.D.McKinney)
Fri Dec 6 00:26:58 2002
From: "W.D.McKinney" <deem@wdm.com>
To: <nanog@merit.edu>
Date: Thu, 5 Dec 2002 20:26:01 -0900
In-Reply-To: <Pine.GSO.4.44.0211202257410.19201-100000@clifden.donelan.com>
Errors-To: owner-nanog-outgoing@merit.edu
Thanks for posting Sean. Any other papers along the same vein ?
Dee
-----Original Message-----
From: owner-nanog@merit.edu [mailto:owner-nanog@merit.edu]On Behalf Of
Sean Donelan
Sent: Wednesday, November 20, 2002 7:17 PM
To: nanog@merit.edu
Subject: Network integrity and non-random removal of nodes
On 20 Nov 2002, William Waites wrote:
> If you randomly select nodes to remove, by the time you have removed
> 25% of them, the network breaks up into many isolated islands.
One of the key points was the nodes were removed in ranked order, not
in random order. Removing the nodes in ranked order result in a linear
decrease in connectivity, i.e. remove the top 1% of the core nodes removes
1% of the connections. But then the scary academic language appears "the
curves appear to be highly asymmetric around a critical point." That is
an understatement like "Houston, we have a problem."
http://www.caida.org/outreach/papers/2001/OSD/
Its a very interesting paper, and I recommend anyone responsible for
network integrity or reliability read it.
