VM Sizing Recommendations
StratoLens analyzes performance metrics from your Azure VMs and VM Scale Sets and recommends smaller, cheaper SKUs when resources are oversized for their actual workload.
What You'll Learn
This documentation covers how to find and act on VM sizing recommendations:
- How recommendations are generated and what feeds them
- Browse, filter, and tune recommendations across subscriptions
- Compare alternative SKUs (cross-family, older generation, B-series) instantly
- Track realized savings after a VM is downsized
- Configure aggressiveness and the minimum data window
Key Capabilities
Quantified Savings Per VM
Every recommendation includes monthly savings, annual savings, and percent reduction so you can prioritize the biggest wins first.
Performance-Based Sizing
Recommendations use observed CPU, memory, and disk metrics over a rolling window of up to 365 days, not guesswork or template defaults.
Instant Alternative SKUs
Toggle cross-family, older generation, B-series, and feature-change options to see different recommendations without re-running a scan.
Confidence Per Recommendation
Each recommendation carries a confidence level (Very High, High, Medium, Low) based on data sufficiency, workload stability, headroom, pattern, and memory data availability.
Realized Savings Tracking
When you downsize a VM, the savings are recorded and visible in the Show Resolved view, including cumulative savings across multi-step downsizes.
Related Features
Feature Integration
VM Sizing works closely with several other StratoLens features:
- Orphaned Resources — VM Sizing only recommends downsizing. For VMs and other resources that appear unused entirely, use Orphaned Resources.
- Performance Metrics — Inspect the underlying CPU, memory, and disk charts that fed a recommendation.
- Resource Explorer — Open any recommended resource in Explorer for full configuration, RBAC, and change history.
- Automated Scanning — Recommendations are regenerated on each scan. Aggressiveness and data-window settings live in Scanner Settings.
Documentation Sections
Start with how recommendations work, then jump into the page itself or the Scanner Settings that control them.
How VM Sizing Works
The recommendation lifecycle, what data feeds it, and the key concepts (workload profile, confidence, aggressiveness, realized savings).
Read: How VM Sizing Works →Recommendations Page
Browse the 3-column hierarchy, filter and toggle alternative SKUs, read recommendation cards, hide noise, and track resolved savings.
Read: Recommendations Page →Scanner Settings
Choose Conservative, Balanced, or Aggressive sizing, and set how many days of metrics a VM needs before its first recommendation.
Read: Scanner Settings →Want to learn more about what VM Sizing Recommendations can do?
Check out the feature page for benefits, use cases, and highlights.
View Feature Page