RE: Definition of Zero Day Protection

["Drew Copley" <dcopley () eEye com>]

Apart from semantical differences over the term "host based", there are
a wide range of heuristic security applications which provide some
degree of protection from zero day.

We have, for instance, long used a "class based" system, in SecureIIS,
which we have greatly expanded in Blink. We have further added multiple
api gating layers and are continuing to greatly expand in this
direction. 

Systrace is an example, among many, of api protection systems. There are
many products in this class. Most of them have limited but realistic
effectiveness against unknown vulnerabilities. How? They limited their
potential destructive influence.

In fact, one of our researcher's [now former] did a presentation at
Black Hat on breaking some of these systems (Seattle). He showed how a
payload could take over a process and spawn new threads, creating an
effective sniffer and trojan agent which by all appearances to most api
protection systems would be the invaded process -- iis.

Regardless, these systems remain our best direction for complete
protection. The hardest trick is not in hardening the system -- it is in
allowing the system to be completely hardened and regulated and to have
it still be usable.

Heuristic AV has long been in the running, though, and many if not most
implementations have detection properties for zero day attacks. AV
generally will not be designed to detect all attacks. The malformed
packet coming in, might not be detected, the resulting shell code may
be. But, the webpage, email, or IM is very likely to be detected. 

Heuristic AV has many problems, however. It is "work in progress". I
made such an agent -- it profiled binaries by apis they used and certain
signatures, such as those for encrypted or packed binaries. Effectively,
I was trying to do what I did manually. And, to some success. The
reasoning is rather simple, if you look at your most common trojan and
malware agents and look for the commonality there. Granted, many virii,
unfortunately, do not have any such common api traits... and it is
always possible not to use typical apis or apis at all to cause damage.

BTW, I mentioned "class based systems". What is that? Ultimately, it
fits in with the "commonality" I was just mentioning. There are certain
commonalities we can find in shell code, in virii, in trojans. I like to
call them "chokepoints", and I like to "gate" these chokepoints. 

For instance, spyware. A vast majority of spyware uses the BHO registry
key. Many use the run registry key on top of that. One can harden these
keys and typically detect and therefore eliminate every spyware which
attempts to use either of these keys -- they are rare enough outside of
the malware world that one might do this.

There are many such chokepoints or commonalities to be found which can
be used as a guide. The trick is to reduce false positives and keep the
system usable. 

**FYI, I will be unable to answer replies, no offense intended to anyone
that might do this. I believe this post was comprehensive.




> -----Original Message-----
> From: Teicher, Mark (Mark) [mailto:teicher () avaya com] 
> Sent: Monday, August 09, 2004 12:15 PM
> To: Drew Simonis; focus-ids () securityfocus com
> Cc: Seanor, Joseph (Joe)
> Subject: RE: Definition of Zero Day Protection
>
> Drew,
>
> What host based products would fit this category based on the 
> definition
> ??  Do they really work ??
>
> -----Original Message-----
> From: Drew Simonis [mailto:simonis () myself com] 
> Sent: Monday, August 09, 2004 01:07 PM
> To: Teicher, Mark (Mark); focus-ids () securityfocus com
> Cc: Seanor, Joseph (Joe)
> Subject: Re: Definition of Zero Day Protection
>
>
> ----- Original Message -----
> From: "Teicher, Mark (Mark)" 
> Date: Sun, 8 Aug 2004 19:47:48 -0600
> Subject: Definition of Zero Day Protection 
>
> > What is Zero Day Protection
>
> It is, as you stated, another marketing blurb, but it isn't just that.
> Usually, this bit of jargon is applied to a 
> detection/prevention system
> that uses things like heuristic detection techniques, behavior based
> detection, protocol anomoly or some other advanced methods.  
> These allow
> the activity to be blocked or alerted on, as opposed to the specific
> event.  
>
> So, for example, a worm can be characterized by certain 
> activity.  Say,
> opening connections to lots of remote hosts in a short period of time.
> This behavior can be blocked (e.g. the process can be killed) even
> without knowing that it was WormX.  
>
>
> hth,
> -Ds
>
>
>
> --------------------------------------------------------------
> ------------
> Test Your IDS
>
> Is your IDS deployed correctly?
> Find out quickly and easily by testing it with real-world 
> attacks from CORE
> IMPACT.
> Go to 
> http://www.securityfocus.com/sponsor/CoreSecurity_focus-ids_04
> 0708 to learn more.
> --------------------------------------------------------------
> ------------

--------------------------------------------------------------------------
Test Your IDS

Is your IDS deployed correctly?
Find out quickly and easily by testing it with real-world attacks from CORE
IMPACT.
Go to http://www.securityfocus.com/sponsor/CoreSecurity_focus-ids_040708 to learn more.
--------------------------------------------------------------------------