A few days ago, I was on the phone with a potential customer who was calling to see what FireMon could do to help them better manage their firewalls. Their particular concern was the performance impact their policy was having on their firewalls. Shortly into the call, they shared the information that their PIX firewall policy contained over 84,000 access control entries (rules). Are you kidding me? Of course performance was a problem, but what about security. This is not a “wall”, at best it is a screen door. I tried to refocus the conversation to security, but performance was their primary concern. So, I started thinking how I could better communicate the particular security concern with a policy permitting so much access.
Firewalls are designed and implemented to control access between networks. Modern-era firewalls are designed with a “positive security model”, simply meaning that they are designed to deny all access that is not administratively permitted. This makes adding rules in a firewall a decision to permit more access. It must also mean accepting some additional risk.
Taking a slightly deeper look into access and risk, it is clear that there is some relationship. At the most basic level, it is understood that the most secure host or network is one where there is no access (think CIA computer in a room with tightly controlled physical access and no network access). Of course there are significant usability issues in such a scenario. So, it also seems clear that permitting more access also increases the risk. Exactly how much will clearly be a factor of the threat posed by the the connecting network, but needless to say, it will increase to some degree. This relationship can be visualized in the below graph. No access presents no risk. A sharp increase in risk once any access is granted and a continual rise in risk as more and more access is permitted.
So, as it relates to the firewall, every rule that permits access also increases risk to some degree. It then seems obvious that excessive access (access that is not needed for any intended purpose) is unnecessary risk. (Clearly removing this excessive access represents low-hanging fruit of risk reduction opportunities.)
However, obviously the customer I was talking to, and in fact many organizations, have focused their firewall management activities on permitting access, not controlling risk. Focus needs to get back to evaluating risk versus access to make the firewall a more effective security device in the network.