This set of metrics allows you to track the processor, memory, and storage usage at the device level. That device can be a physical machine or a virtual machine.
First, make sure you configured the Portal settings to store the new data as by default it is disabled.
The new activities are available in the Finder. You can use them in investigations as well as incorporate them in custom metrics. Also available in the Portal, you can use them in dashboards.
All the data associated with the device performance can be queried using NXQL. To learn more about NXQL, follow our tutorial.
The Collector reports sampled values every minute to the Engine. The data is computed every 15 minutes.
Average CPU usage
The average CPU usage reports the average processing capabilities of a virtual environment. That value can be higher than 100%. It corresponds to the total of all the processors.
A system with 4 CPUs will return an average CPU usage of 200% when 2 CPUs are measured at 20% usage and 2 others at 80% usage.
An average CPU usage of 400% would mean that the processing power on that system might be suffering a lack of resources as intensive tasks are using 100% of each CPU.
Normalized CPU usage
The normalized CPU usage reports how much of the available processing power is utilized. That value cannot be higher than 100%. It is calculated by dividing the total CPU usage by the number of logical CPUs.
A system with 4 CPUs will return a normalized CPU usage of 50% when 2 CPUs are measured at 20% usage and 2 others at 80% usage.
(20% + 20% + 80% +80%) / 4 = 50%
A normalized CPU usage value below 10% means the system is over-provisioned whereas a value above 90% means the system is under-provisioned.
Average memory usage
The average memory usage reports the average memory usage of the device across all device performance events.
Average storage read/write operations
The average read operations per second reports the average number of disk read operations per second across all device performance events.
The average write operations per second reports the average number of disk write operations per second across all device performance events.
Average CPU queue length
The average CPU queue length reports the average CPU queue length across all device performance events. It indicates the number of threads waiting for a CPU to be available.