Great Circle Associates Firewalls
(March 1995)
 

Indexed By Date: [Previous] [Next] Indexed By Thread: [Previous] [Next]

Subject: Re: FW-1, etc.
From: patrick @ oes . amdahl . com (Patrick Horgan)
Date: Fri, 3 Mar 1995 08:45:29 +0800
To: patrick @ oes . amdahl . com, horn @ mickey . jsc . nasa . gov
Cc: firewalls @ GreatCircle . com

> 
> Suppose that you have a user who goes on travel and wants to telnet in to his
> machine FOO from remote host BAR.

The easiest way to fix this is by policy;)  No incoming connections.

>                                    To allow this, the packet filtering
> firewall must allow in port 23 from BAR to FOO.  Now, there are two ways to
> get around this type of authentication:
> 
> 	a) IP address spoofing - i.e. pretend that you are host BAR
> 	b) Get access to host BAR, and telnet in
> 
> The latter of the two is particularly nasty because it's probably not that
> difficult.  Using a packet filter, and IP based authentication, when you
> trust host BAR you also implicitly trust all of the hosts that BAR trusts,
> and all that those hosts trust, etc, etc.  Breaking into any one of the hosts
> in that web of trust means access to your inside machines.
> 
> The biggest problem with packet filter based firewalls is the basic trust in
> IP addresses for authentication.  Now, I don't know anything particular about
> FW-1, although I've been told several times that it handles UDP in a
> reasonable fashion (OK, so call me skeptical).  Nevertheless, without strong
> USER authentication at the firewall, it isn't anything more than a fancy GUI
> packet filter, which is inherently less secure than an application level
> gateway.

So what if you had something that listened on port 23 on the firewall, did
reasonable authentication either via an encrypted channel with exchange of
public keys, or via a single key method.  Once authenticated you'd be
able to back proxy to where you wanted to go.  This, I believe is a good
way to go, there's a real synergy between things like this and a "I hate
to say just packet filter, since FW-1 does some application stuff," packet
filter.

> 
> Of course, if your packet filter is one where absolutely NO inbound traffic is
> allowed, and you trust your internal users, then I'd be inclined to say that a
> packet filter is no less secure than an app-gateway.  However, as soon as you
> want to allow inbound traffic, the game is up, and the security of the packet
> filter breaks down.


I'd like a combination of free outgoing via the packet filter, and incoming
via some sort of authenticated channels.

Nothing here shows any hole implicit in FW-1 though.  (You know it's funny
that FW-1 is most often complemented for their GUI in reviews, but I found
it completely non-intuitive.  I thought it was one of the worst designed
user interfaces...just trying to figure out how to apply a change to a 
filter was bizarre.)

Patrick
   _______________________________________________________________________
  /  These opinions are mine, and not Amdahl's (except by coincidence;).  \
 |                                                       (\                |
 |  Patrick J. Horgan         Amdahl Corporation          \\    Have       |
 |  patrick @
 amdahl .
 com        1250 East Arques Avenue      \\  _ Sword     | 
 |  Phone : (408)992-2779     P.O. Box 3470 M/S 316         \\/    Will    | 
 |  FAX   : (408)773-0833     Sunnyvale, CA 94088-3470     _/\\     Travel | 
  \___________________________O16-2294________________________\)__________/

Indexed By Date Previous: Re: set group id on directories
From: patrick @ oes . amdahl . com (Patrick Horgan)
Next: IPX traffic through a firewall
From: stagda @ sys1 . ic . ncs . com (Dave Stagner)
Indexed By Thread Previous: Re: FW-1, etc.
From: "Marcus J. Ranum" <mjr @ tis . com>
Next: Re: FW-1, etc.
From: Brent @ GreatCircle . COM (Brent Chapman)

Google
 
Search Internet Search www.greatcircle.com