Hi Akamas,
Based on this part of the documentation:
https://www.ibm.com/docs/en/instana-observability?topic=technologies-monitoring-cri-o#ariaid-title11
The underlying source value cpuTotalUsage represents a percentage (ratio) calculation derived from nanoseconds.
The Linux kernel (via cgroups) tracks container processing in raw time counters called cpuTotalUsageNanoseconds. To make this actionable, the Instana agent evaluates these counters over a specific time window.
cpuTotalUsage = (cpuTotalUsageNanoseconds / systemTime)
- cpuTotalUsageNanoseconds: The total amount of CPU time (in nanoseconds) used by the container's processes during that monitoring time window.
- systemTime: The total wall-clock system time available during that exact same window.
It's a unitless ratio (decimal value) and Instana multiplies this ratio by 100 and formats it as a Percentage.
Now, the real question, How to interpret this number?
The value represents how many complete CPU cores the container fully occupied during that interval.
- If the UI reads
50% (0.5), the container used exactly half of one CPU core's processing capacity over that time window.
- If the UI reads
100% (1.0), the container completely saturated exactly one CPU core.
- If the UI reads
250% (2.5), the workload was distributed, consuming the equivalent power of (2.5) fully taxed CPU cores.
And that's it, this is my understanding for that metrics and I hope it helps you.