Kubernetes - Forecasted saturation recommendation

The Forecasted saturation recommendation detects entities such as clusters, nodes, or containers for which the resource usage has already exceeded thresholds or is expected to exceed the thresholds within the next 30 days. The recommendation is generated by evaluating the natural (organic) growth of resource usage in your entities due to workload growth or deployment of new projects (cluster only). It is based on the historical trending and extrapolation to the future. It compares the estimated resource utilization with thresholds.

You can configure a recommendation as per your requirement by modifying the condition in the corresponding Optimizer rule. A user with administrative privileges can reconfigure or add new Optimizer rules to modify the recommendations. For more information, see Configuring and managing Optimizer rules.

The following recommendation details are provided:

  • Number of days that are remaining for the resources saturation
  • Severity or Efficiency level of the recommendation. The thresholds for calculating the efficiency level is defined in the associated optimizer rule.
  • Date when this recommendation is generated on. It is the date of the last run of the associated optimizer rule.
  • Name of the associated optimizer rule
  • Recommended actions to resolve the resource saturation
  • Actions that you can perform on your recommendations. For details, see Managing the recommendations.
  • Comparison between the current resource usage and trend and the estimated future usage after reconfiguring the system
  • Criteria used for generating the recommendation
  • Indicator thresholds used for the recommendation
  • Benefits of implementing the recommendation

Forecasted saturation for infrastructure entities such as clusters, namespaces, or nodes

This recommendation rule is based on the following thresholds:

  • CPU Utilization 
  • Memory Utilization
  • Storage Utilization

Forecasted saturation for containers

This recommendation rule is based on the following thresholds:

  • CPU Utilization based on CPU requests
  • Memory Utilization based on Memory requests

The warning threshold values that are configured for these metrics are used for the recommendation.  You can configure them on the Thresholds page in the Administration > Optimizer section.

For information about configuring thresholds, see Configuring and managing thresholds for metrics and indicators.

The recommendation state is marked as critical if the resource saturation is expected to occur within the next 7 days or warning if it is expected to occur within the next 8 to 30 days. You can change these default values on the Rules page in the Administration > Optimizer section. For information about modifying the Optimizer rules, see Configuring and managing Optimizer rules.

The minimum 5 days of data is required to calculate the utilization trend of resources. You can modify this setting on the indicators page in the Administration > Optimizer section. For details, see Configuring and managing thresholds for metrics and indicators.

Was this page helpful? Yes No Submitting... Thank you