Since every customer environment varies, it is recommended to gather baseline production statistics over a set period of time in order to see if there are any outliers during the baseline period. During this baseline, note periods when client requests/second where maximized. If there was a corresponding drastic increase in response time and potential CPU usage, this could represent the peak performance in this specific environment. Further testing and monitoring should be performed in order to confirm this maximum level.
After the baseline period has elapsed, and no specific maximum peaks have been observed in client requests /second, it is recommend to artificially set a threshold value of 10% to 25% of the highest observed client requests/second for alerting purposes.
Aside from monitoring performance and alerting on specific exceeded thresholds, the Cisco WSA can also be configured to send SNMP traps on these hardware conditions: