Dataverse Storage Capacity: Database, File and Log Limits, Costs and Cleanup (2026)

Last updated: November 13, 2026

By ZapAI Team

Every tenant that runs Power Apps, Power Automate or Dynamics 365 gets a pooled allowance of Dataverse database, file and log storage, and it’s the first number our Dataverse services team asks for in an admin review. Go over it and Microsoft blocks new environments, copies and restores. Database capacity lists at $40 per GB per month. File costs $2. Log costs $10.

Capacity warnings used to be easy to ignore. The banner nagged, the weekly email landed in a shared inbox, and nothing stopped working. Not anymore. Microsoft began rolling out storage validation on November 9, 2026, and a Dataverse-only tenant that stays over capacity now watches its sandboxes move into administration mode after 30 days and get disabled after 90.

This guide covers what fills each capacity type, what your licenses include, where the admin center reports live, what overage triggers, what extra capacity costs, and how to clean up without deleting something an auditor needs. License basics sit in our Power Platform licensing guide. We won’t repeat them. Every product fact here was checked against Microsoft Learn and the November 2026 licensing guides on November 13, 2026.

What Is Dataverse Storage Capacity?

It’s your tenant’s licensed allowance of storage in three separate buckets, called database, file and log. Each bucket has its own entitlement, its own add-on and its own price. Usage from every counted environment draws on the same tenant pool.

Microsoft’s Dataverse capacity-based storage page spells out which tables land where. Short version below.

A few rules catch people out.

• Trial, preview, support, developer and Teams environments don’t count. They show as 0 GB.
• Each counted environment consumes at least 1 GB, with or without a database, according to Microsoft’s capacity FAQ.
• The default environment carries its own included 3 GB of database, 3 GB of file and 1 GB of log.
• Finance and operations customers see Dataverse and Operations capacity listed separately. Database and file still pool for enforcement.

One reporting quirk. In April 2026 Microsoft moved customization metadata from database to file reporting, so that month’s database dip was bookkeeping. Not cleanup.

How Much Capacity Do Your Licenses Include?

Your first qualifying subscription brings a one-time tenant default. Most full user licenses then add a slice of database and file capacity per user. Log mostly doesn’t grow with headcount.

Dynamics 365 rows come from the November 2026 Dynamics 365 Licensing Guide, which says plainly that default capacity isn’t cumulative. Extra base or attach licenses don’t raise the default, and attach licenses carry no capacity of their own, Customer Insights aside. Growth comes from accrual, pooled tenant-wide.

One wrinkle. The Power Platform guide’s table lists a 20 GB database default for Power Apps Premium, while the worked example a page later still starts a new Power Apps tenant at 10 GB. Its change log says Microsoft raised Power Platform defaults in November 2025, so we read the table as current. Your admin center has the final word, under See Dataverse capacity per license.

Here’s a hypothetical example. Say a tenant started on Customer Service Enterprise with 40 agents and later added 15 Power Apps Premium makers. Database works out to 30 GB plus 10 GB plus 3.75 GB, or 43.75 GB. File comes to 150 GB. Log stays at 2 GB. That last number bites, because for these licenses log only grows when you buy log.

Seller tier differences are in our Dynamics 365 Sales pricing breakdown, and Customer Service Premium’s 35 GB file accrual, meant for call recordings per the licensing guide, gets weighed in our Dynamics 365 Customer Service guide.

Where Do You Check Capacity in the Power Platform Admin Center?

Go to Licensing, then Dataverse under Products. That page shows tenant usage by storage type, the environments consuming the most, a per-table view for each environment, and the Manage capacity panel. The Capacity add-ons page still has Summary, Dataverse, Microsoft Teams, Add-ons and Trial tabs.

Access matters. Only tenant admins, Power Platform admins and Dynamics 365 admins see the tenant Summary, and environment admins get redirected to the environment view, so check a capacity owner’s admin role before anyone opens a ticket. Bought Dynamics 365 through a partner? Microsoft says to contact that partner to manage capacity.

Reports also lag. Storage is evaluated about every 24 hours, deletions can take up to 72 hours to show, and the warning banner refreshes on its own seven-day cycle, so a banner can linger for days after the problem is fixed.

What Happens When You Exceed Dataverse Capacity?

First warnings, then blocks. Admins get banners and weekly emails below 15% remaining, below 5% and over 100%, and an over-capacity tenant can’t create, copy, restore, recover or convert environments. Dataverse-only tenants now face a sandbox lifecycle on top.

Before any of that, Microsoft applies cross capacity-type borrowing. Spare database capacity can cover log or file overuse. Spare log can cover file. Nothing covers a database shortfall.

Careful, though. Notifications use the borrowed position, but environment operations check the native capacity type, so Microsoft says an operation can be blocked even when the tenant report shows no deficit after borrowing. Read both numbers.

Scope is narrow but real. The validation covers Dataverse-only tenants running Sales, Customer Service, Field Service, Customer Insights, Customer Voice, Contact Center or Project Operations on Dataverse, and in those tenants production environments don’t enter administration mode or get disabled. Neither do sandboxes linked to pay-as-you-go. Tenants with finance and operations environments sit outside it for now.

Message center post MC1483519 lists the rollout as November 9 to December 14, 2026. One third-party summary we found said sandboxes are disabled after 60 days, while Microsoft’s Learn page and the message center both say 90. Plan around 90. When the overage clears, environments don’t switch back on their own, so an admin has to exit administration mode or re-enable each one.

There’s also a short bridge. Once consumption passes 80%, a tenant admin can turn on a Dataverse capacity extension of 25% for 45 days, up to three times in 365 days. It’s temporary. Use it to finish a purchase order. Not to postpone cleanup.

What Does Extra Dataverse Capacity Cost?

At list, prepaid add-ons run $40 per GB per month for database, $2 for file and $10 for log, billed annually. Pay-as-you-go meters charge $48, $2.40 and $12. Agreement pricing can differ, so ask your reseller for your rate.

Prices come from the November 2026 Power Platform Licensing Guide and Microsoft’s pay-as-you-go meters page. The public Power Apps pricing page shows only the database add-on.

Look at the ratio. At list, a GB of database costs 20 times a GB of file. Moving data can beat deleting it, and buying file capacity to cover a database deficit does nothing, since Microsoft says file capacity can’t resolve a database deficit, no matter how much of it you own.

Pay-as-you-go deserves its own note. Link an environment to an Azure billing policy and its usage above the included 1 GB of database and 1 GB of file bills to Azure, measured daily, instead of drawing on the tenant pool. That suits a spiky sandbox. Sustained growth costs less on prepaid add-ons, $40 against $48 per database GB.

Do Copilot, Agents and Managed Environments Use Dataverse Capacity?

Copilot features can. Files you upload to a Copilot Studio agent are stored in Dataverse along with their search indexes, and the Dataverse search index itself bills at the database rate, the priciest of the three. Managed Environments matter for another reason. Long-term retention needs one.

Microsoft’s Copilot Studio unstructured data page draws a clean line. The SharePoint option under file upload copies files into Dataverse and consumes storage for the copies and their indexes. The SharePoint connector option leaves content in SharePoint and consumes none.

Search is the quieter cost. The DataverseSearch table holds the index behind global search, generative answers and Copilot semantic indexes. Don’t switch Dataverse search off to save space. Microsoft warns that dependent AI features stop working, and after 12 hours the index is deleted and has to rebuild. Trim indexed tables and columns instead.

Link to Fabric belongs here too. Its Delta Parquet replica counts toward Dataverse database capacity, which our Dataverse Link to Fabric guide sizes in detail. Agent credits are a separate meter, covered in our Copilot Studio pricing guide.

What’s Eating My Dataverse Capacity?

Usually a handful of system tables, not your business data. Open the environment’s Consumption per table view, sort by size, and match the top tables against this list before you buy anything.

That last row trips up careful admins. With Keep deleted Dataverse records turned on, deleted rows stay for the 1 to 90 days you set and keep counting toward capacity until they’re purged, per Microsoft’s deleted records page. Bulk deletion jobs have a Permanent deletion checkbox that skips that holding step.

AI Builder document processing models also keep training documents in Dataverse file and image columns. Unused models still take space.

How Do You Reduce Dataverse Storage Usage?

Fix the cheapest growth first. Turn off plug-in tracing, set audit retention, let system job cleanup run, bulk delete or retain old activities and attachments, then remove environments nobody uses. Microsoft’s free up storage space guide lists 15 methods. These move the numbers.

Log capacity, audit and trace logs

Log is where small tenants hit the wall. Two GB doesn’t grow with users. Two settings do the heavy lifting.

1. Set an audit retention period. Microsoft’s auditing guide lists Forever as the default, and changing it from the environment settings only applies to new logs, while the Security page option can apply it to existing ones. Agree the period with compliance first.
2. Delete old audit logs by table, by access log, or by date from the environment’s Auditing card. Microsoft puts the rate at about 100 million records a day.
3. Set plug-in trace logging to Off in production. Microsoft’s logging and tracing page says trace logs take up storage and should run only while debugging, a daily job deletes trace records older than 24 hours, and Microsoft might disable logging if the table passes 100 GB.

System jobs and workflow logs

1. Check the deletion service. By default it removes succeeded system jobs after 30 days and failed or canceled ones after 60, per Microsoft’s system job cleanup page. A preview setting under Audit and logs exposes those periods.
2. Review waiting workflows. Cleanup only removes completed jobs. Some wait on purpose.
3. Open each classic background workflow and turn on Automatically delete completed workflow jobs.

Activities and attachments

Bulk deletion lives under the environment’s Settings, then Data management, then Bulk deletion, as Microsoft’s bulk deletion guide shows. Microsoft’s own examples are recurring jobs that clear sent or received email older than a month with attachments over 1 MB, plus notes carrying files over 1 MB, each scheduled to run outside working hours. Pick your own thresholds. One month is aggressive.

Moving files out is the alternative. Server-based SharePoint integration keeps new documents in SharePoint libraries instead of Dataverse, the setup our Dataverse vs SharePoint Lists guide recommends for documents. It doesn’t move existing notes, though, so for those you need a partner tool or custom code that copies attachments to SharePoint or Azure Blob storage and leaves a link behind.

Long-term retention for data you must keep

Retention moves inactive rows into a compressed, read-only store inside Dataverse, and Microsoft’s long-term retention overview says each GB moved there uses on average 50% less database capacity than it did as live data. Mind the fine print.

• The environment must be a Managed Environment.
• Retained data can’t come back.
• Attachments get no savings.
• Audit and elastic tables aren’t supported.
• Runs take 72 to 96 hours.
• Policy runs consume Power Platform requests.

Managed Environments licensing is covered in our Power Platform governance guide.

Fabric is an analytics copy, not an archive

Fabric frees nothing by itself. Link to Fabric adds a replica to database capacity, and Fabric only becomes an archive when you copy data out, confirm it, and then delete or retain the source rows in Dataverse, which is a design call our Microsoft Fabric team makes table by table.

Environments

Each counted environment holds at least 1 GB, and a sandbox copied from production carries the whole database with it, which is how a few forgotten copies can push a tenant over without anyone adding a single record. Delete sandboxes nobody touches, after checking owners and backups. Deleted environments stay recoverable for seven days. Our dual-write vs virtual tables comparison explains why dual-write sandboxes multiply storage.

Should You Clean Up or Buy More Capacity?

Clean up database, buy file, fix log settings. That’s our default rule. Compare a year of add-on cost for the GBs at stake with the one-time cost of cleaning them up, and clean up when payback lands inside about three months.

Take a worked example with made-up numbers. A tenant is 25 GB over on database, which at the $40 list price is $1,000 a month, or $12,000 a year, and cleanup takes 16 hours of admin time at an assumed internal cost of $150 an hour, so $2,400 that pays back in under three months. Now run the same 25 GB on file. The add-on costs $50 a month. Payback takes four years. Buy the file capacity.

For the purchase itself, borrow Microsoft’s sizing formula. Required capacity equals the effective deficit plus expected growth plus an operating buffer, minus confirmed cleanup that hasn’t shown up in reports yet.

What Should a Monthly Capacity Review Look Like?

Thirty minutes. One owner. Same day each month. This is ZapAI practice, not a Microsoft rule.

1. Open Licensing, then Dataverse.
2. Note database, file and log use against entitlement in a running log, so you see growth per month instead of just today’s total, which is the only reliable way to catch a table that doubles between quarterly reviews.
3. Open the top three environments.
4. Compare each top table to last month, and treat anything growing faster than the business as a question for its owner.
5. Confirm tracing is Off.
6. Check that recurring bulk deletion jobs ran.
7. Review idle sandboxes with their owners.
8. Forecast the months left until 85% at the current growth rate, and start a purchase now if the answer is under three.

The forecast step is the one teams skip. Procurement takes weeks. Storage validation won’t wait.

What Mistakes Push Tenants Over Capacity?

Small defaults left alone. In our experience the same handful keep showing up.

• Plug-in tracing left on All.
• Audit retention left at Forever.
• Several full sandbox copies of production.
• File capacity bought for a database deficit.
• Bulk deleting with deleted-record keeping on.
• Uploading whole SharePoint libraries to Copilot Studio agents.
• Reallocating capacity between environments to fix a tenant overage, which Microsoft says adds no capacity.
• Living on capacity extensions.

Migrations deserve their own warning. Ten years of email attachments that cost nothing on premises land straight in file capacity online, a point our CRM on-premises to online migration guide makes too. Trim before you move.

Dataverse Capacity Checklist

Run it once now, then monthly.

Know your numbers

• Entitlement by type, from the license breakdown.
• Current use and growth.
• Your reseller’s price per GB.
• Whether you’re Dataverse-only.

Fix the settings

• Plug-in tracing Off in production.
• Audit retention approved by compliance.
• Deletion service on.
• Workflow job auto-delete on.
• Short deleted-records window.
• Dataverse search trimmed.

Clean and protect

• Recurring bulk delete jobs scheduled.
• Retention policies for inactive data.
• New documents in SharePoint.
• Idle sandboxes deleted.
• Environment alerts set in Manage capacity.

What Could Change After Ignite 2026?

Maybe the numbers. Ignite runs November 17 to 20, 2026, in San Francisco, and Microsoft has changed Dataverse defaults more than once since November 2025. We’ll recheck entitlements, the storage advisor’s status and the validation rollout by late November.

Wrapping Up

Dataverse capacity is three meters, not one. Database is the expensive one and can’t borrow. File is cheap. Log is tiny and doesn’t grow with users. With storage validation rolling out, an ignored overage now costs sandbox access in Dataverse-only tenants, so the monthly check earns its half hour.

ZapAI is a Microsoft Dynamics 365, Power Platform, Fabric and AI consulting firm in Irving, Texas, working with businesses across Dallas-Fort Worth and nationwide. Our Dataverse consultants size capacity, set up cleanup jobs and retention policies, and decide what belongs in Dataverse versus SharePoint or Fabric. Already over? Ask us for a Dataverse capacity review, and we’ll show you what’s eating your storage before you buy more.

Questions Admins Ask About Dataverse Capacity

How much Dataverse storage capacity do we get by default?

What’s the difference between database, file and log capacity?

What happens if we go over Dataverse capacity?

How much does extra Dataverse storage cost?

Does deleting records free up capacity right away?

Do Copilot Studio agents use Dataverse storage?

Does long-term retention reduce file capacity?