The tool and the research that went into its creation are the work of Ivan Ristic, director of engineering at security vendor Qualys and the original author of the popular ModSecurity Web application firewall.
Web application firewalls are designed to protect Web applications from known attacks, such as SQL injection attacks, that are commonly used to compromise websites. They do this by intercepting requests sent by clients and enforcing strict rules about their formatting and payload.
However, there are various methods for sneaking malicious requests that violate these rules past WAFs by modifying certain parts of their headers or the paths of requested URLs. These are known as protocol-level evasion techniques, and WAFs are not properly equipped to deal with them at the moment because the techniques are not very well documented, Ristic said.
The researcher tested the evasion techniques he found primarily against ModSecurity, an open source Web application firewall, but it’s reasonable to assume that other WAFs are vulnerable to some of them as well.
In fact, Ristic said he shared a few of the techniques with others during the research stage and that they had tested them successfully against some commercial WAF products.
Erwin Huber Dohner, head of research and development at Switzerland-based WAF vendor Ergon Informatik, confirmed after seeing Ristic’s presentation that the evasion methods are a problem for the industry. Ergon recently identified some similar techniques that worked against its product and have addressed them, he said.
By making his research public, Ristic hopes to kick start a discussion in the industry about protocol-level and other types of evasion. A wiki has also been set up with the purpose of building a freely available catalogue of WAF evasion techniques.
If vendors and security researchers don’t document the problems and make them known, WAF developers will make the same mistakes over and over, Ristic said.
In addition, the availability of the testing tool will allow users to discover which WAF products are vulnerable and hopefully force vendors to fix them.
Vendors have different priorities and don’t normally fix things unless there’s a real risk to their customers, Ristic said. This research project will hopefully generate the necessary incentive for them to deal with these issues, he said.
Dohner welcomed the initiative and believes that it will benefit WAF developers and users alike.