We are using EC2 instances and are monitoring the server usage.
Right now the main metric we are looking at is idle CPU: Monitoring load on AWS EC2
However, I am also hoping to monitor "load" as measured by the top command. We are running on large instances, so it appears to us as if we have 2 cores.
For fully owned servers, I usually set the trigger value to slightly higher than the number of cores (e.g. if 2 cores, trigger an alert if the load is over 2.5 for more than a minute). Is this still the right thing to do on EC2?
After finding out that load measures correctly for each user on EC2, we went with our standard load triggers.
For us that is half a CPU above the max for more than 5 minutes.