What Microsoft announced
Azure Storage Discovery provides enterprise-wide visibility into storage usage, activity, egress, and configuration across your data estate. The new preview builds on those estate-wide insights with a more granular view of how data is stored and used. The Microsoft Learn documentation says the Capacity details report complements Capacity overview. The overview shows how much storage you have, and the details show how you're consuming it.
There is one important boundary. The insights aggregate object metadata at the storage-account, container and configured-prefix levels. They do not provide an inventory of individual blobs. This is a triage tool. It is not a blob-by-blob audit.
Cost and prerequisites
- Microsoft's blog says advanced capacity insights are included in the Azure Storage Discovery Standard plan at no additional cost.
- Microsoft's planning documentation requires an existing Storage Discovery workspace on the Standard pricing plan. The feature isn't available with the Free plan.
- During preview, you enable the feature after workspace creation.
- The planning documentation lists the permissions needed to enable it. They are Contributor or Owner on the workspace, plus advanced capacity access permissions on the workspace's root entries.
- Allow up to 48 hours for initial data to populate before you rely on the report.
What the report shows
The report has several parts, each answering a different question.
- Size by tier and recency. This compares capacity or object count across Hot, Cool, Cold and Archive. You can view it by creation time, last-accessed time or last-modified time. The age bands run from 0–3 days up to more than 365 days. It can surface old data sitting in a hot tier, or busy data parked in a cold one.
- Size-bucket distribution. This splits capacity and object count into five ranges, from 0–1 KB to more than 1 TB. A high share of tiny objects points to a workload that needs a closer look. A few huge buckets show you what dominates your footprint.
- Minimum-retention analysis. This flags Cool, Cold or Archive data still inside its minimum-retention window. Those windows are 30, 90 and 180 days respectively. Deleting, overwriting, moving or retiering data early can trigger a prorated early-deletion charge.
- Storage composition and soft-deleted retention. These compare live data against previous versions, snapshots and soft-deleted data. They also show how long soft-deleted data stays recoverable and keeps counting toward billable capacity.
- Top 10 items. This ranks accounts, containers or prefixes by size or count. Depending on the options available, rankings can be top, bottom, fastest growing or fastest shrinking. Microsoft cautions that growth rankings need enough report history to mean anything.
How to open and use it
- In the Azure portal, open your Storage Discovery workspace.
- Select Reports, then Capacity details.
- Choose the Discovery scope you want to analyze.
- Apply filters. Microsoft lists subscriptions, accounts, containers and configured prefixes. It also lists region, performance type, encryption type, redundancy type, tags and time range.
- On any chart, select View resources to see the accounts, containers or prefixes behind a segment.
- Use Edit columns to choose fields. Use Download as CSV to export.
Microsoft warns that grouping in the portal may look different in the CSV. Check the exported columns before feeding the file into automation.
Regional availability
Availability has moved quickly. Some early coverage described a small initial footprint, but the current Microsoft Learn table is much broader. It lists regions available since October 22, 2026, such as Canada Central, Canada East, East Asia and France Central. It lists many more available since November 2, 2026, including East US, West Europe, Japan East and UK South.
Microsoft says coverage can still be partial within a listed region. If insights are missing for an account, wait a few days, then contact Azure Support if the problem persists. Check the live table for your own regions rather than relying on any single announcement.
Preview limits that affect your numbers
| Area | Preview coverage |
|---|---|
| Storage accounts | Up to 50,000 per Discovery scope, prioritized by size |
| Objects per account | 10 billion (flat namespace), 5 billion (hierarchical namespace) |
| Container-level insights | Top 100 containers by size per included account |
| Configured prefixes | Up to 100 per workspace |
Accounts over the object limits may be left out. Container totals may cover only some containers. Microsoft says the report shows a coverage message when the whole estate isn't included, and you should read it before comparing totals across levels.
Last-access analysis only works for accounts with last-access-time tracking enabled. If some accounts lack it, those results can be incomplete. Microsoft points to the Configurations report to check this.
Practical advice
Treat the findings as leads. Microsoft says plainly that the report is for investigation. Review workload requirements, lifecycle policies and current Azure Storage pricing before changing or deleting anything.
An independent commentary post, Beyond Cloud with Chriz, makes a similar point from the operations side. It suggests getting data-owner approval before turning a capacity trend into a lifecycle change. It also warns against converting an apparent share of inactive data into a promised saving. Soft-deleted data and versions often exist for recovery reasons, so the lost capability should be weighed alongside the capacity saved.
My own read, based on general industry experience rather than Microsoft's text, is that the retention-exposure chart is the most useful part. It helps you avoid early-deletion charges while you clean up. The small-object view is a close second. Many tiering and transaction-cost surprises start with huge counts of tiny blobs.
Bottom line
Preview means non-production use and testing, so don't build critical automation on it yet. For Blob Storage administrators and FinOps teams on the Standard plan, it is a free way to see where capacity is going. Enable it, give it two days, check the coverage message, and treat what it shows as leads to investigate.