Firewall Rule Best Practices¶
This section covers general best practices for firewall rule configuration.
There are two basic philosophies in computer security related to access control: default allow and default deny. A default deny strategy for firewall rules is the best practice. Firewall administrators should configure rules to permit only the bare minimum required traffic for the needs of a network, and let the remaining traffic drop with the default deny rule built into pfSense. In following this methodology, the number of deny rules in a ruleset will be minimal. They still have a place for some uses, but will be minimized in most environments by following a default deny strategy.
In a default two-interface LAN and WAN configuration, pfSense utilizes default deny on the WAN and default allow on the LAN. Everything inbound from the Internet is denied, and everything out to the Internet from the LAN is permitted. All home grade routers use this methodology, as do all similar open source projects and most similar commercial offerings. It’s what most people expect out of the box, therefore it is the default configuration. That said, while it is a convenient way to start, it is not the recommended means of long-term operation.
pfSense users often ask “What bad things should I block?” but that is the wrong question as it applies to a default allow methodology. Noted security professional Marcus Ranum includes default permit in his “Six Dumbest Ideas in Computer Security” paper, which is recommended reading for any security professional. Permit only what a network requires and avoid leaving the default allow all rule on the LAN and adding block rules for “bad things” above the permit rule.
Keep it short¶
The shorter a ruleset, the easier it is to manage. Long rulesets are difficult to work with, increase the chances of human error, tend to become overly permissive, and are significantly more difficult to audit. Utilize aliases to keep the ruleset as short as possible.
Review Firewall Rules¶
We recommend a manual review of the firewall rules and NAT configuration on a periodic basis to ensure they still match the minimum requirements of the current network environment. The recommended frequency of such reviews varies from one environment to another. In networks that do not change frequently, with a small number of firewall administrators and good change control procedures, quarterly or semi-annually is usually adequate. For fast changing environments or those with poor change control and several people with firewall access, review the configuration at least on a monthly basis.
Quite often when reviewing rules with customers we ask about specific rules and they respond with “We removed that server six months ago.” If something else would have taken over the same internal IP address as the previous server, then traffic would have been allowed to the new server that may not have been intended.
Document The Configuration¶
In all but the smallest networks, it can be hard to recall what is configured where and why. We always recommend using the Description field in firewall and NAT rules to document the purpose of the rules. In larger or more complex deployments, create and maintain a more detailed configuration document describing the entire pfSense configuration. When reviewing the firewall configuration in the future, this will help determine which rules are necessary and why they are there. This also applies to any other area of the configuration.
It is also important to keep this document up to date. When performing periodic configuration reviews, also review this document to ensure it remains up-to-date with the current configuration. Ensure this document is updated whenever configuration changes are made.
Reducing Log Noise¶
By default, pfSense will log packets blocked by the default deny rule. This means all of the noise getting blocked from the Internet will be logged. Sometimes there will not be much noise in the logs, but in many environments there will inevitably be something incessantly spamming the logs.
On networks using large broadcast domains – a practice commonly employed by cable ISPs – this is most often NetBIOS broadcasts from clue-deficient individuals who connect Windows machines directly to their broadband connections. These machines will constantly pump out broadcast requests for network browsing, among other things. ISP routing protocol packets may also be visible, or router redundancy protocols such as VRRP or HSRP. In co-location environments such as data centers, a combination of all of those things may be present.
Because there is no value in knowing that the firewall blocked 14 million NetBIOS broadcasts in the past day, and that noise could be covering up logs that are important, it is a good idea to add a block rule on the WAN interface for repeated noise traffic. By adding a block rule without logging enabled on the WAN interface, this traffic will still be blocked, but no longer fill the logs.
The rule shown in Figure Firewall Rule to Prevent Logging Broadcasts is configured on a test system where the “WAN” is on an internal LAN behind an edge firewall. To get rid of the log noise to see the things of interest, we added this rule to block – but not log – anything with the destination of the broadcast address of that subnet.
We recommend adding similar rules, matching the specifics of any log noise observed in an environment. Check the firewall logs under Status > System Logs, Firewall tab to see what kind of traffic the firewall is blocking, and review how often it appears in the log. If any particular traffic is consistently being logged more than 5 times a minute, and the traffic is not malicious or noteworthy, add a block rule for it to reduce log noise.
Out of the box, pfSense does not log any passed traffic and logs all dropped traffic. This is the typical default behavior of almost every open source and commercial firewall. It is the most practical, as logging all passed traffic is rarely desirable due to the load and log levels generated. This methodology is a bit backwards, however, from a security perspective. Blocked traffic cannot harm a network so its log value is limited, while traffic that gets passed could be very important log information to have if a system is compromised. After eliminating any useless block noise as described in the previous section, the remainder is of some value for trend analysis purposes. If significantly more or less log volume than usual is observed, it is probably good to investigate the nature of the logged traffic. OSSEC, an open source host-based intrusion detection system (IDS), is one system that can gather logs from pfSense via syslog and alert based on log volume abnormalities.