IDS mailing list archives

RE: interesting paper on testing sig-based IDS


From: "Brian Smith" <bsmith () tippingpoint com>
Date: Thu, 3 Mar 2005 11:03:12 -0600

For what it's worth, I've seen IPSs that cannot deal with TCP or IP fragmentation or overlap testing, so it's worth 
testing it.  Fragroute is the standard tool for doing this testing.  Whisker and ADMutate are also good off the shelf 
tools for evasion testing.  I believe the latest version of Blade's software has evasion testing built in.  When using 
these tools, make sure that the set of evasions your testing don't neuter the attack.  It's possible to use these tools 
to hide an attack so effectively with evasions that they evade the target system!
 
The other thing you may have to consider when testing inline IPSs is to realize that they become part of the network 
infrastructure -- every packet in the network goes through them, often multiple times.  So evaluating performance is 
often as important as evaluating seucrity, and the actual performance of the devices is all over the map.  If their 
throughput, latency, or other performance characteristics are unacceptable in your situation, that's also a show 
stopper.  We have a customer, for instance, that regularly processes > 600K packets per second.  This was a requirement 
for his environment, and if an IPS couldn't keep up with this, it wasn't worth considering.
 
The tricky part of this testing performance is that the code path executed by an IPS, and therefore the performance, is 
dependent on the type of traffic.  For example, a different code path will be executed for HTTP vs. DNS.   If 
processing a DNS packet takes a long time, this won't show up if you test with a pure HTTP mix, or a smartbits set to 
send ethernet frames filled with 0s.  So it's important to use the protocol mix used in the target network when testing 
performance. Tomahawk (tomahawk.sourceforge.net) is a tool I developed to replay traffic from a target network through 
the IPS for performance testing.
 
Finally, test that the IPS can block attacks and maintain throughput at the same time.  It doesn't do much good if the 
IPS starts missing attacks under load, or slows to a crawl under attack.  I usually test with an attack rate I'd expect 
in the next worm outbreak (maybe 100 - 2000 attack/second, depending on how fast your network is).
 
HTH
 
            Brian

________________________________________
From: buineach [mailto:securesolutions () gmail com]
Sent: Tue 3/1/2005 5:58 PM
To: Jonathon Giffin
Cc: Kohlenberg, Toby; focus-ids () lists securityfocus com; Shai Rubin
Subject: Re: interesting paper on testing sig-based IDS
Hi
I just joined this forum so apologies if this has been asked/answered before.

Is this tool available to the general public as I do a lot of IPS
testing and would like to verify further the framentation and TCP
segment handling of these inline products. ?
I have been assuming that all current IPS products have mechanisms to
deal with evasion techniques like this but as the NSS testing results
show a lot of current IPS solutions are nothing more than the offline
IDS they were before with many signatures disabled with 2 NIC's.

A real concern I have with inline IPS that depend on a central CPU to
deal with fragmentation and segmentation evasion is that an overload
attack with this traffic will make the IPS the weakest link in the
network.
I have ruled out many IPS vendors based on using ISIC through the IPS
but would like to have a more specific tool to deal with TCP segment
shifting with metasploit framework for example to see who fails here.

Any info appreciated.

Mick




On Mon, 28 Feb 2005 10:32:20 -0600, Jonathon Giffin <giffin () cs wisc edu> wrote:
Kohlenberg, Toby wrote:
http://www.cs.ucsb.edu/~vigna/pub/2004_vigna_robertson_balzarotti_CCS04.
pdf

You may also be interested in Automatic Generation and Analysis of NIDS
Attacks by Rubin, Jha, and Miller from ACSAC 2004.

http://www.cs.wisc.edu/wisa/papers/acsac04/RJM04.pdf

Abstract:

A common way to elude a signature-based NIDS is to transform an attack
instance that the NIDS recognizes into another instance that it misses.
For example, to avoid matching the attack payload to a NIDS signature,
attackers split the payload into seversl TCP packets or hide it between
benign messages. We observe that different attack instances can be
derived from each other using simple transformations. We model these
transformations as inference rules in a natural-deduction system.
Starting from an exemplary attack instance, we use an inference engine
to automatically generate all possible instances derived by a set of
rules. The result is a simple yet powerful tool capable of both
generating attack instances for NIDS testing and determining whether a
given sequence of packets is an attack.

In several testing phases using different sets of rules, our tool
exposed serious vulnerabilities in Snort--a widely deployed NIDS.
Attackers acquainted with these vulnerabilities would have been able to
construct instances that elude Snort for any TCP-based attack, any
Web-CGI attack, and any attack whose signature is a certain type of
regular expression.

Disclaimer: I am part of the same research group as the authors of this
paper.

Thanks,

Jon

--------------------------------------------------------------------------
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.
--------------------------------------------------------------------------



--------------------------------------------------------------------------
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.
--------------------------------------------------------------------------

--------------------------------------------------------------------------
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.
--------------------------------------------------------------------------


Current thread: