Good question. No real easy answer to it. Lot of things involved. I would
definetly say, "Do NOT allow just any traffic to go out!" Specify which
machines will be allowed to host a web server. Block everything going out
that you do not specifically let out. It would also make people use a proxy
server to surf the web, not just go out. 

Chris
Keeper of the digital flame.

-----Original Message-----
From: Ferry van Steen [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, April 23, 2002 6:19 AM
To: [EMAIL PROTECTED]
Subject: How to get through iptables/NAT, reality and risk calculation


Hey there,

first of all, please don't get me wrong. I don't want to know how to crack a
firewall, I just don't wanna think I'm secure whilst I'm not.

The case is this, at several locations I've set up a linux box for the
internet traffic. These boxes are configured in such a way that they don't
have any open ports (or atleast, not on the internet side). This is
accomplished by simply allowing all traffic from the local LAN but only
accepting traffic from the internet part of an existing connection (with the
iptables -m state --state ESTABLISHED,RELATED).

Now, to me, as starting security engineer (security-guru-wannabe or whatever
the phrase is), this looks uncrackable to me (unless people download and
install trojans that connect to IRC n stuff, which is allowed (atleast,
according to traffic rules :-))). What should I be aware of? Could people
for instance get data into the network by hiking along on a connection
somebody set up with a webserver (or any other service for that matter)? The
people on these locations are allowed to do whatever they want, they can
IRC, MSN, ICQ, HTTP, HTTPS, etc... Would it be possible that the linux box
gets hacked due to a TCP/IP stack bug? I'm just sucking things out of my
thumb here so I hope they make sense. Every knowledgeable security engineer
I ever spoke say nothing is uncrackable, so I'm just trying to figure out
the ways they still can get it so I can do things to prevent those and/or
atleast analyse the risk and have a knowledge of the possibilities so I
won't be utterly suprised somewhere in the future without a clue as to where
to look and how to trace it back.

I'm really sorry if this has been discussed before... The site is really
slow at the moment. In any case all info is welcomed (URLs, books,
references, user stories, experiences... whatever).

Btw.. I'm subscribed to the list on another email addy than this one. I am
subscribed tho'. Replying to either this email ([EMAIL PROTECTED])
or the list would be fine.

Kind regards and TIA,

Ferry van Steen

Reply via email to