# Workload Recommendations

The Workload Recommendations page provides a comprehensive view of optimization opportunities across all your Kubernetes workloads. Use this page to identify which workloads can be rightsized, review the overall impact of applying recommendations, and prioritize changes based on potential savings or risk reduction.

![Workload Recommendations Table](/files/dUiK1BbRlu93uQDQ4s1w)

## Filtering the Analysis

The analysis can be filtered by cluster and namespace to focus on specific parts of your infrastructure. Use these filters to narrow down results when working with large environments or when targeting specific teams or applications.

## Impact Summary

At the top of the page, summary cards show the aggregate impact of applying all displayed recommendations.

### Efficiency Impact

The efficiency cards display the total resource changes if all recommendations are applied:

* **CPU**: Current vs. recommended total CPU requests across all workloads;
* **Memory**: Current vs. recommended total memory requests across all workloads;
* **Cost Savings**: Estimated monthly savings from rightsizing.

### Reliability Impact

The reliability cards show how recommendations affect workload stability:

* **Risks Before**: Count of reliability risks in the current configuration;
* **Risks After**: Count of risks after applying recommendations;
* **New Risks**: Any potential risks introduced by tighter resource limits.

## Recommendations Table

The table lists all workloads with their current configuration, detected opportunities, and recommended changes.

### Key Information

Each row displays:

* **Workload Name**: Click to open the detailed workload analysis page;
* **Namespace and Type**: Location and kind of workload (Deployment, StatefulSet, etc.);
* **Current Resources**: Existing CPU and memory requests and limits;
* **Recommended Resources**: Suggested values based on observed usage;
* **Opportunities**: Color-coded badges indicating optimization opportunities;
* **Savings**: Potential monthly cost reduction for this workload.

### Opportunity Categories

Opportunities are organized into three categories, each represented by color-coded badges:

* **Efficiency** (green): Over-provisioned resources where requests exceed actual usage;
* **Reliability** (blue): Configuration risks such as high throttling, OOMKill risk, or missing limits;
* **Best Practices** (purple): Violations of Kubernetes configuration standards.

Some recommendations may introduce new potential risks (blue badges), indicating that tighter limits could cause throttling or memory pressure under peak load. Review these carefully before implementation.

## HPA Recommendations

Workloads managed by a Horizontal Pod Autoscaler (HPA) receive a different type of recommendation. Instead of per-container CPU/memory sizing, the platform recommends optimal **maxReplicas** and **minReplicas** values based on observed peak and baseline demand.

### How HPA Workloads Appear

* **HPA column**: The recommendations table shows an HPA status badge (Detected / Not Detected) for each workload;
* **Replicas column**: Shows average and peak replica counts;
* **Workload detail page**: Displays an HPA metadata card (current min/max replicas, target utilization) and HPA-specific recommendation cards.

### HPA Findings

The platform detects several HPA-related findings:

* **"HPA max replicas reached"** (Reliability): The workload hit its configured maximum, potentially unable to handle peak load;
* **"HPA never scales down from max replicas"** (Efficiency): The HPA ceiling may be set too low if the workload is always at max;
* **"HPA rarely scales down from max replicas"** (Efficiency): Similar, but the workload occasionally scales down;
* **"HPA min replicas equals max replicas"** (Compliance): HPA cannot autoscale when min == max;
* **"Workload could benefit from HPA"** (Efficiency): CPU demand is spiky but no HPA is configured.

## Next Steps

Click any workload name to view detailed charts, usage patterns, and specific implementation instructions. See [Applying Recommendations](/insights/applying-changes/applying-recommendations.md) for guidance on implementing changes safely.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.akamas.io/insights/analysis-and-recommendations/workload-recommendations.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.