Orion uses a couple of methods to monitor up/down status and to collect statistical data (cpu, memory utilization, bandwidth utilization, etc.) Under the Network Performance Monitor Settings, you have a Polling tab and a Statistics tab.
The Polling tab refers to your up/down status. That is monitored using a ping that is sent out every x (x being your polling interval.) In your case, it sounds like you are pinging your server every minute. That gets you up/down status but won't get you the statistical data like cpu utilization.
The Statistics tab is where you configure the default statistics collection period for your nodes. That data is collected via SNMP polls of your devices. Typically, this is done less frequently than the up/down pings.
In your case, you would need to increase the statistics collection interval for your server to a minute in order to be alerted if the CPU utilization is at 100%. You are correct in assuming that the utilization has to be at 100% when the server is polled in order for Orion to know it is over the alert threshold.
Hope this helps. Let me know if you have any other questions.
Thanks a lot. That does help.