OWASP AppSec 2010: Automated vs. Manual Security: You Can't Filter The Stupid 1/3

Clip 1/3
Speakers: David Byrne and Charles Henderson, Trustwave

Everyone wants to stretch their security budget, and automated application security tools are an appealing choice for doing so. However, manual security testing isn't going anywhere until the HAL application scanner comes online. This presentation will use often humorous, real-world examples to illustrate the relative strengths and weaknesses of automated solutions and manual techniques.

Automated tools certainly have some strengths (namely low incremental cost, detecting simple vulnerabilities, and performing highly repetitive tasks). In addition to preventing some attacks, WAFs also have advantages for some compliance frameworks. However, automated solutions are far from perfect. To begin with, there are entire classes of very important vulnerabilities that are theoretically impossible for automated software to detect (at least until HAL comes online). Examples include complex information leakage, race conditions, logic flaws, design flaws, subjective vulnerabilities such as CSRF, and multistage process attacks.

Beyond that, there are many vulnerabilities that are too complicated or obscure to practically detect with an automated tool. Automated tools are designed to cover common application designs and platforms. Applications using an unusual layout or components will not be thoroughly protected by automated tools. Realistically, only the most vanilla of web applications written on common, simple platforms will receive solid code coverage from an automated tool.

On the other hand, manual testing is far more versatile. An experienced penetration tester can identify complicated vulnerabilities in the same way that an attacker does. Specific, real-world examples of vulnerabilities only recognizable by humans will be provided. The diversity of vulnerabilities shown will clearly demonstrate that all applications have the potential for significant vulnerabilities not detectable by automated tools.

Manual source code reviews present even more benefits by identifying vulnerabilities that require access to source code. Examples include "hidden" or unused application components, SQL injection with no evidence in the response, exotic injection attacks (e.g. mainframe session attacks), vulnerabilities in back-end systems, and intentional backdoors. Many organizations assume that this type of vulnerability is not a large threat, but source code can be obtained by disgruntled developers, by internal attackers when the repository isn't properly secured, by exploiting platform bugs or path directory traversal attacks, and by external attackers using a Trojan horse or similar technique.

For more information click here (http://bit.ly/aeSvg2)

Category:

Science & Technology

Tags:

• SebastienDeleersnyder
• Day1
• Track2
• 15301605860
• clip1