[31260] in bugtraq
Re: Buffer overflow prevention
daemon@ATHENA.MIT.EDU (Mark Handley)
Mon Aug 18 16:38:56 2003
From: Mark Handley <M.Handley@cs.ucl.ac.uk>
To: Crispin Cowan <crispin@immunix.com>
Cc: "BUGTRAQ@SECURITYFOCUS.COM" <BUGTRAQ@securityfocus.com>
In-reply-to: Your message of "Sun, 17 Aug 2003 15:42:07 PDT."
<3F4004BF.10108@immunix.com>
Date: Mon, 18 Aug 2003 19:07:07 +0100
Message-ID: <17684.1061230027@aardvark.cs.ucl.ac.uk>
>Heterogeneity increases survivability of the *species*, but does little
>to protect the individual.
What you're not taking into account is contagion. Amongst a
homogeneous population, a pathogen that infects your friends can
likely infect you. Amongst a heterogeneous population, if the same
pathogen infects a friend, there's a significantly lower probability
it can infect you.
Now, if you're promiscuous and come into contact with enough
strangers, you'll catch the pathogen either way. But if you're not
promiscuous, you greatly reduce the change of contracting the pathogen
if you are part of a heterogeneous population.
How does this affect networks? Well, if you're a webserver or
mailserver that talks to everyone, the heterogeneity doesn't buy you
so much (other than, as you said, there might be more pathogens for
popular systems). But if you're configured to not talk to the whole
world (via a firewall, or something equivalent), then you're a whole
lot safer if the machines you do communicate with are different from
you in ways that make contagion harder.
Cheers,
Mark
