Using the Performance Metrics Page
Open the Performance Metrics page from Performance in the left sidebar, between Access and Network Visualizer. Use it to find a specific Azure resource and read its real utilization pattern over the past 1, 7, 14, or 30 days.
On This Page
Use the 95th percentile for rightsizing decisions
A VM running at a 95th percentile CPU of 25% can typically move to a smaller SKU. One running at 90% cannot. The 95th percentile filters out brief spikes that pull the peak up and rare lulls that pull the average down, leaving the value the resource actually sustains.
Finding a Resource
The page opens with three columns: Subscriptions, Resource Groups, and Resources. Drill left to right.
- Click a subscription in column 1. Column 2 fills with its resource groups.
- Click a resource group in column 2. Column 3 fills with its supported resources.
- Click a resource in column 3. The detail panel on the right populates with the resource's metrics.
Each row in the first two columns shows the name and a resource count. Rows in column 3 show the resource type icon, the resource name, and the resource type string (for example, microsoft.compute/virtualmachines).
Collapse columns once you've drilled in
After you select a subscription or resource group, a < chevron appears in that column's header. Click it to collapse the column to a thin vertical strip and free up horizontal space for the chart. Click again to expand.
Reading the Detail Panel
The detail panel header shows the resource name, its resource type, and, for Virtual Machines, the VM size (for example, Standard_D8s_v3) after a bullet separator. Below the header is a row of metric tabs and, beneath the active tab, a summary card and chart.
Metric Tabs
Tabs are generated from what Azure Monitor returns for the resource type. A VM typically shows CPU Utilization, Memory Available, Disk IOPS, and similar; a Cosmos DB account shows Request Units and Total Requests; and so on. The Collection Window tab is always the last tab.
Switching tabs replaces the summary card and chart for the selected metric. Switching to a different resource resets the active tab to the first available metric for that resource type.
Summary Card
The summary card sits between the tabs and the chart. It surfaces up to four numbers in a 2×2 grid:
- Average
- Mean of all data points in the selected time window.
- 95th Percentile ★
- The value the resource runs at or below for 95% of the window. The recommended signal for rightsizing.
- Minimum
- Lowest observed value, with a relative timestamp like "3 days ago". Hover for the exact local time.
- Peak
- Highest observed value, with a relative timestamp.
Some metrics show only Average and 95th Percentile
Metrics that Azure Monitor reports as a single aggregation (for example, Total Requests is Total-only, Session Count is Average-only) hide the Minimum and Peak row. The 2×2 collapses to a single row.
Reading the Chart
The chart sits below the summary card and renders the selected metric across the active time window. It does not appear on the Collection Window tab.
Lines
Which lines are drawn depends on the metric:
- Multi-line metrics
- Three lines: Minimum (dotted, green), Average (solid, blue), Maximum (dashed, red). Used for CPU, Memory, Disk IOPS on VMs and similar.
- Total-only metrics
- Single solid blue Total line. Examples: Request Units (RU), Total Requests.
- Average-only metrics
- Single solid blue Average line. Examples: Session Count, Storage Usage.
- 95th Percentile ★
- Orange/amber dashed horizontal reference line, always drawn, labeled with its value (for example,
95th %ile: 24%) on the right edge.
The 95th percentile line cannot be hidden
The Show metrics dropdown above the chart toggles Minimum, Average, and Maximum on or off. The 95th percentile reference line stays visible regardless.
Hover Tooltip
Hover any point to see the formatted date, the raw date string, and each visible metric value with its color. The 95th percentile sits at the bottom of the tooltip below a divider so it's easy to scan against the live values.
Gaps in the Chart
Missing data shows as a gap rather than a zero. Hover the gap to see why:
- Resource may be powered off
- Gap inside the collected window. The resource was deallocated or stopped during that interval.
- Awaiting next scan
- Gap after the most recent scan timestamp, typically at the right edge of a freshly loaded chart.
Filters and Time Window
The detail panel header carries two dropdowns on the right: time window and granularity. A third dropdown sits above the chart for memory metrics on VMs, and a Show metrics dropdown lets you hide individual lines on multi-line charts.
Defaults
- Time window
- Past 7 Days (options: Past 1 Day, Past 7 Days, Past 14 Days, Past 30 Days)
- Granularity
- Daily (options: Daily, Hourly)
- Memory display
- Used (GB) (options: Used (GB), Available (GB), Used (%), Available (%))
- Active metric tab
- First available metric for the resource type, typically CPU Utilization for compute resources.
Time window and granularity persist in the URL
Both are written to the URL (for example, ?granularity=hourly&window=14), so the browser Back button undoes filter changes and you can paste a link to share a specific zoom level.
When the Granularity Dropdown Is Hidden
The Daily / Hourly dropdown only appears when both granularities exist for the resource. Some resource types (Cosmos DB, NAT Gateway, Traffic Manager, Storage Accounts, Azure Bastion) only have hourly data, so daily charts for them look flat or aggregated. Switch to hourly for the real shape.
Memory Display Mode
On a memory metric for a VM, a dropdown above the chart switches between Used (GB), Available (GB), Used (%), and Available (%). StratoLens stores the raw Available Memory Bytes value Azure reports and converts in the browser, so swapping modes is instant.
Collection Window Tab
The Collection Window tab (always the last tab) shows how much history StratoLens has on file for this specific resource. It's the answer to "why does my chart only go back two weeks?" on a brand-new resource.
- Daily Metrics
- Number of days, plus a date range like
2025-04-25 to 2026-04-24. - Hourly Metrics
- Number of hours, plus a date range. Only shown when hourly data exists for the resource.
New resources have shorter windows until they've accumulated more scans. To check or change how often scans run, see Automated Scanning.
Troubleshooting
My resource isn't in the list
What to check
Either the resource type isn't in the supported list above, the most recent scan hasn't completed, or Azure Monitor returned no data and StratoLens is still building history. Run a scan from Scan Overview and check again once it finishes.
My VM looks idle but the chart is empty
What's happening
A deallocated VM emits no metrics, and StratoLens doesn't store all-zero rows. The Collection Window tab still shows the date range, but the metric chart will appear blank or sparse.