What you’re seeing
Some dashboards or alerts may display CPU load average instead of CPU utilization percentage. These metrics are related, but they do not measure the same thing.
What is load average?
Load average represents the average number of processes that are runnable or waiting for resources over a period of time. Systems commonly report load averages for the previous 1, 5, and 15 minutes.
Load average should be interpreted in relation to the number of CPU cores:
- On a single-core system, a load average of 1.0 indicates that the CPU was fully loaded on average.
- On a four-core system, a load average of 1.0 represents approximately 25% of the total CPU capacity.
- A load average above the number of available CPU cores generally indicates sustained contention or queuing.
What is CPU utilization percentage?
CPU utilization percentage measures the percentage of CPU time spent processing work instead of remaining idle during a sampling interval.
For example:
- 20% CPU utilization means the CPU was active for approximately 20% of the sampling period.
- 90% CPU utilization indicates that the CPU was busy for most of the sampling period.
Why the values may differ
Load average includes processes waiting for CPU time or other resources, such as disk or network I/O. As a result:
- Load average may be high while CPU utilization is moderate if processes are waiting for I/O.
- CPU utilization may be high while the load average remains relatively low during a short burst of activity.
- A load average must be evaluated against the number of CPU cores before it can be interpreted.
Which metric should I use?
Use load average for:
- Identifying sustained workload or process queuing.
- Comparing workload trends over time.
- Understanding system pressure beyond CPU activity alone.
Use CPU utilization percentage for:
- Identifying immediate CPU saturation.
- Monitoring firewall, router, or appliance performance.
- Creating alerts based on a device’s available processing capacity.
For firewalls and network appliances, vendor-specific CPU utilization metrics are generally more useful for determining whether the device is approaching capacity.
If CPU utilization is not available
The device may expose CPU utilization through its management interface, CLI, or vendor-specific SNMP objects. Ask your network administrator to confirm whether the device provides a CPU utilization metric and whether the required monitoring credentials and OIDs are available to Auvik.
The available metric depends on the vendor, device model, firmware, and monitoring method.