Performance monitoring eases maintaining mission-critical services by not only alerting you when the service goes down but also informing you when the service performance degrades. Degraded performance can cause a bad user experience, or it even might be foreshadowing a future downtime.
Let's say your service generally responds in 300 milliseconds, and a higher response time requires attention. But getting notified of every single response with a response time greater than 300 milliseconds would be very noisy. Connection hiccups or momentary issues would cause pointless notifications. But if the service keeps responding late for a particular time, then most probably, there is really an issue.
Enabling performance monitoring
You can enable performance monitoring on the HTTP monitor's settings.
There are 3 parameters that need to be set:
- Number of responses required to mark monitor as slow
- The time window for checking response times
- Response time limit
Getting alerted when monitor is slow
Alerts have specific configuration options for slow, just like downtime. Please see alerts documentation for details.