Firewall policies are complex. Some firewall vendors try and reduce the complexity of administrating firewalls with graphical editors or zone-based administration and those concepts help. However, the reality of legacy policies, short timelines for engineering changes and staff reductions all lead to the strong likelihood that a large number of technical errors exist in firewall policies and more are being made.
Realizing this gives us a great starting point for detoxing the firewall. Let’s clean up the technical errors — those configuration elements that could never be accessed based on the policy above. Hidden rules, obscured rules, covered up objects…all are errors in the policy that can be removed. One of the reasons that I like to start a firewall cleanup with technical errors is that there is no need to collect logs or consult the business because removing those items does not change the behavior of the firewall.
The first challenge is finding these errors. In a policy with even a couple hundred rules, the errors can be difficult to find manually. Automating the detection of these items is a good starting point and allows us to continue to monitor for errors after the initial cleanup is complete. FireMon’s Hidden Rules Report was built for this purpose.
Once we find the errors, we need to prioritize the remediation. Some errors are worse (more obvious and usually easier to clean up) than others, and there is the possibility of many errors in a current policy. Traditionally, we consider rules that are completely hidden by the policy above to be the starting point for removing errors. It’s the most obvious place to start and it can have the most impact. After we resolve the completely hidden rules, we can work our way toward other error conditions, like covered services or network objects. The Hidden Rules Report offers results in graduated levels of detail so we can see precisely the level of information we need. No more, no less.
Removing technical errors is just the first step towards detoxing your firewall.