The RSA show is around the corner, and undoubtedly next-generation firewalls will be a hot conversation topic. These firewalls represent a big leap forward in firewall technology and the timing is perfect; Web 2.0 applications have grown, and simply controlling ports and protocols is an insufficient defense of increasingly sophisticated attacks.
These firewalls are able to control access based on application behavior, permitting a user to browse the Web while limiting Web-based file sharing, even though both may transfer data on TCP port 80. These solutions advocate controlling access by user instead of by IP address. This method is a much more dynamic and potentially effective way to control access — outbound access in particular.
The new capabilities also encourage a new approach to firewall policy management. Instead of static IP address and protocol/port definitions in a rule, an administrator may use user groups and application categories as part of the rule definition. This method permits a one-time configuration setup, while enabling the configuration to dynamically adapt to changes in user groups and application definitions.
However, these solutions are not without their own perils. While traditional firewall management may be less dynamic, it is mostly self-contained. So, an administrator can definitively state what access is permitted based on the policy definition, and then evaluate changes to the policy to understand the impacts to the permitted access.
This simple evaluation is not so simple with dynamic policies. Changes to Active Directory may have unintended implications for firewall behavior. Updates to the dynamic application filters may inadvertently block access to a critical business application or grant access to an application that should be blocked.
An even more basic problem is the conflicts between rules that result in failures. Although the next-generation firewalls are unique in their ability to filter applications by behavior, they still follow the traditional firewall paradigm of top-down filtering wherein the first applicable rule is applied. Managing applications may be preferred, but often, legacy applications or custom applications require management of protocol and port access. As a result, it is still possible to inadvertently allow or deny access — resulting in hidden and shadowed rules — in even more ways with next-generation firewalls.
I don’t see any of these challenges as problems with the next-generation firewalls. They are simply reminders that like most security technology, next-generation firewalls are only as good as their configurations. Failure to effectively manage the configuration will cripple the security capability of this powerful new security solution.