Policy performance collection configuration
The following example shows how to activate the policy performance collection function. Policy performance data for all active policies is maintained by the TCP/IP stack. This function allows this data to be collected and made available for policy performance monitoring applications. The following statements apply to the example in this topic:
- Performance data is collected for both rules and actions. Action data is an aggregate of the data for the rules that refer to the action. For policies that have a single action per rule, the performance data will be the same for both the rule and action.
- The default minimum sampling interval is 30 seconds. This is the minimum value accepted for the acceptableCachedTime parameter on the PAPI papi_get_perf_data() function that gets performance data.
- Performance data will be logged to the file /u/user10/perflog, based on a sampling interval of 60 seconds.
- The number of performance log files maintained is 10, each of
which is the default 300 kilobytes in size. In this example, the log
files will be named assuming a stack name of TCPCS:
/u/user10/perflog.TCPCS /u/user10/perflog.TCPCS.1 /u/user10/perflog.TCPCS.2 ⋮ /u/user10/perflog.TCPCS.9
- Each performance data record is 232 bytes, so a file size of 300
kilobytes can contain 1324 records. Because 10 files are maintained,
the total set of log files can contain 13240 records. Assuming that
50 policy rules and actions exist in the configuration, this means
that the set of log files will wrap in approximately 4.4 hours, according
to the following formula:
number of records (13240) / number of policies (50) = number of refresh cycles (264) number of refresh cycles (264) * refresh interval in minutes (1) = 264 minutes worth of data
PolicyPerformanceCollection Enable
{
DataCollection Rule Action
LogSamplingInterval 60
PerformanceLogFile /u/user10/perflog
NumberOfLogFiles 10
}