Exception profiling fine tunes security policies associated with a Service to reduce wrongly blocked requests. Typically, when a Service is created, it is associated with the“default security policy. All URLs and Parameters are compared to the default security policy settings, which may block some genuine requests. Blocked genuine requests are called false positives. Exception Profiling uses a heuristics based strategy to refine web application security settings in response to logged traffic in the Web Firewall logs, viewable on BASIC > Web Firewall Logs. It reduces the false positives by creating URL and Parameter profiles which relax the default security policy settings or by modifying existing URL profile or Parameter profile settings thus fine-tuning them to the Service. If a profile does not exist, Exception Profiling can create a new profile, or if a profile already exists it is fine tuned to reflect the exception.
You can enable Exception Profiling for a Service by clicking Edit next to the service on the WEBSITES > Exception Profiling page and setting the Exception profiling level to one of three levels for untrusted traffic. Each level has corresponding settings for exception handling. The levels are:
Trusted traffic is always handled according to separately configured Trusted settings.
Exception Profiling Levels
A Low Profiling Level indicates a low tolerance to violations, so logged traffic violations are reviewed frequently to properly adjust security settings. On the other hand, a High exception profiling level indicates a greater tolerance of violations, and a higher confidence in the security settings, so review or adjustment of the profile only happens if a violation is seen more frequently. Exception profiling treats traffic violations differently for trusted hosts. Trusted hosts heuristics have a "Trigger Count" permanently set to one (1) for all violations. Any request from a trusted host is assumed to be a genuine request; so an exception from any trusted host automatically refines security, creating or modifying profiles for the Service as required.
Because only three levels of Exception Profiling Heuristics for untrusted traffic apply to all services, a change in the settings of any level applies to any service using that level (Low, Medium, or High). Exception profiling provides default settings for each violation type. The settings indicate how exceptions update profiles (Automatically, Manually, or not at all), how the new setting in the profile is generated (increasing the current value, or accepting the observed value, for example) and how many times the logged error needs to be seen before generating an exception (Trigger Count). These default settings for an Exception Profiling Level can be edited and saved. For information on how to edit request violation settings, see Configuring Exception Heuristics.
Selecting an Exception Profiling Level of Low will increase the number of exceptions or recommendations for profile adjustment, causing a more rapid adjustment of the profile to reflect observed traffic. On the other hand, an Exception Profiling Level of High results in fewer exceptions and pending recommendations, indicating increased confidence in the profile, and higher tolerance for traffic violations. The Trusted profiling level is recommended for trusted hosts.
Configuring Exception Profiling
You can configure exception profiling for a Service by setting the exception profiling level, thereby applying the corresponding Exception Heuristics settings to that Service. Perform the following steps to configure exception profiling:
- From the WEBSITES > Exception Profiling page identify the Service for which you want exception profiling enabled.
- Click Edit next to that Service. The Edit Exception Profiling window appears.
- To learn from a trusted hosts group, select the trusted host group from the Trusted Hosts Group drop-down list and set Learn From Trusted Host Group to Yes. For information on trusted hosts, see .
- To learn from untrusted traffic, select the level of tolerance to violations (Low, Medium or High) from the Exception Profiling Level drop-down list. For information on exception profiling levels, see .
- Click Save.
Configuring Exception Heuristics
The WEBSITES > Exception Heuristics page allows you to view the definitions for any exception profiling level and adjust settings for various Violation Types if required.
Exception Profiling Level Settings
Exception Profiling Level determines the exception creation heuristics for the Service to which it is bound. Four policies, or levels, are provided: Low, Medium, High (for untrusted traffic) and Trusted.
To view the settings for a profiling level:
- Go to the WEBSITES > Exception Heuristics page.
- Select the desired level to view the heuristics settings in the Exception Profiling Level module, and click Show Definition.
The Request Violation Handling module gets populated with the settings for that level and can be modified. The levels are shareable across multiple Services. Any change made to an exception heuristics level setting applies to any Service bound to this level. Services may have an untrusted traffic exception profiling level (Low, Medium, or High) as well as designated trusted hosts using the Trusted Hosts exception profiling settings.
The exception heuristics for various Violation Types are classified into:
- Length Violations
- Input Violations
- Header Violations
- Cookie Violations
- Forceful Browsing
For each violation type, set the following parameters:
- Setting – How exceptions are created: Automatically, Manually through approval of Pending Recommendations, or no exceptions should be created.
- New Value – How to modify the parameter after learning. New Value can be a function of the old value (increase 100% for example). Or the new value can be based on the default option provided. New Value is selected from provided options.
- Trigger Count – This threshold sets the number of times a violation must be received from unique sources before triggering exception learning either automatically or manually. Only unique requests from a client are counted. Multiple violations from the same client generate a single violation in the trigger count. This neutralizes a hacker conducting repeated attacks on the Service. An exception profiling agent processes the web firewall logs which generated the violations defined on the Exception Heuristics page. It maintains a cache of the trigger counts and compares the running count with the configured trigger count. When the trigger count is met, it invokes the exception profiling process.