Min/Max Recommendations

Min/Max Recommendations give you ML-driven minimum and maximum replica boundaries for Autopilot. Wave analyzes each Deployment's traffic and recommends the replica bounds to use, in two flavors: a Performance set (more headroom) and a Cost set (leaner).

How it works

Wave learns a Deployment's traffic pattern (24-hour cycles, over 7 to 14 days of history) and produces two recommended sets of bounds:

• Performance: Min Performance and Max Performance keep more headroom for spikes. Use for production, user-facing services.
• Cost: Min Cost and Max Cost run leaner to save cost. Use for cost-sensitive or latency-tolerant workloads.

There is no separate toggle. Recommendations appear automatically once Autopilot is enabled on the Deployment and enough data has been collected.

Viewing recommendations

1. Enable Autopilot (or Monitoring) on the Deployment; see Getting Started.
2. Open the Deployment and go to its Min/Max Recommendation tab.
3. Review the Performance and Cost cards.

Below the cards, a 24-hour heatmap and area chart show predicted replicas by hour, so you can see peak and off-peak windows.

Applying the values

Copy the recommended values into your Autopilot config:

1. Go to the Deployment's Autopilot settings.
2. Set Min Replicas and Max Replicas from your chosen set (Performance or Cost).
3. Save, and monitor scaling for about a week.

Troubleshooting

Next Steps

• Core Configuration: all Autopilot settings
• Cost vs Performance: choosing a strategy