RE: About Data Control

["Gonzalez, Albert" <albert.gonzalez () eds com>]

Well,

        My honeypot is currently inside of my internal network.
I have it on its own internal private subnet separated from
all my other machines. I have my gateway limit X connections
per X minutes/seconds outbound from my honeypot. Thus if my 
machine does get breached while im not watching (which usually -
happens) the attacker can't do much with the limitations I empose.+
(I've even have had them to the point of rm'ing my machine :( ). 
The Honeynet Project (correct me if im wrong) is trying to use 
snort-inline to block the "stupid" attacks(ie: CodeRed, IIS Unicode, etc.)
and other well known attacks to see if they can get something new. 
Which is a great idea. HTH.

Cheers!
   Alberto Gonzalez. 

-----Original Message-----
From: Martim Carbone [mailto:martim.carbone () ic unicamp br]
Sent: Friday, January 17, 2003 1:16 PM
To: honeypots () securityfocus com
Subject: About Data Control


Hi,

I am currently working on the Data Control part of my Honeynet, and have 
already configured Snort-inline to run with the rc.firewall script 
provided by the Honeynet Project. This configuration could prevent exploit 
attacks, scans and some DoS attacks. However, there is still one type of 
"attack" this setup does not prevent.

Suppose a random attacker breaks into a random machine A on the 
Internet, installs a backdoor and then breaks into OUR honeypot.
He could effectively use our honeypot as a bounce station and 
anonymize his connection to his backdoor on host A. And as far as I know, 
neither snort-inline nor the connection-limiting scheme could prevent 
him from doing it. Needless to say, this  could get the honeynet's
administrators 
into serious trouble if  A's administrators find out where the attacker is 
connecting from.

Any ideas on how to prevent this?

Thanks,

-- Martim