Instant Recommendations are suggestions for fine-tuning the resource utilization of Kubernetes® workloads. They are based on recent historical usage and are derived from up to two weeks of resource utilization data.

Instant Recommendations are ideal for the following use cases:

  • Initial improvements before launching active optimization
  • Improving efficiency and reducing costs for workloads allocated excessive resources
  • Improving performance and reliability for workloads that don't have sufficient resources to operate reliably 
  • Optimizing workloads with multiple containers in the pod that drive resource use and performance

View Instant Recommendations

To view an instant recommendation:

  1. From Observe, scroll to Kubernetes.
  2. Click Workloads.
  3. From the Workloads list view, select Optimize.
  4. Click a workload from the Instant Recommendation column, which will indicate that an instant recommendation is available.
  5. From the Workload detail view, scroll down to the Application Resource Optimization widget.
  6. Select Instant Recommendation.

Anatomy of an Instant Recommendation

The Confidence Level is based on analyzing historical data, patterns, and trends. The confidence level can have the following values:

  • Low 
  • Middle 
  • High
The Cost Impact is calculated as a percentage (Cost Increase) and a dollar amount (Estimated Monthly Increase). The values can be positive or negative. A suggested increase in cost would indicate that more resources are needed for optimal performance. The Estimated Monthly Increase panel is available only if Cost Insights is also subscribed.
The Overall Resource Change provides the suggested number of CPU cores and memory (GiB). 
Resource Specifications provides a YAML snippet that you can copy and paste into the specs.container[] field of your Kubernetes deployment YAML file.

When to Run Active Optimization Tests 

Instant Recommendations are a quick and efficient way to determine the right size of workload resources and are widely accepted as a best practice approach.

ARO also provides Active Optimization for enhanced recommendations for:

  • Workloads with application performance monitoring (APM) metrics and that have understood and important performance objectives

  • Business-critical workloads requiring performance verification before recommendations can be applied

  • Workloads expected to have significant variations of load over time, including expected future changes