Azure DevOps Backup & Recovery
Microsoft protects the platform. You protect the data.
Microsoft replicates your data within a region and to a second region, and keeps point-in-time database backups for 28 days. That covers outages and disasters. It does not cover accidental deletion. Backing up your own data is your responsibility.
Native Recovery Has Limits
Recycle bins exist. They help. But each one works differently, and all of them live inside the same tenant that had the incident.
| What native recovery does | Where it falls short |
|---|---|
| Deleted projects: recoverable for 28 days, then gone for good | No point-in-time restore. Support cannot roll you back to last Tuesday. |
| Deleted repos: 30 days in the recycle bin | Nothing for history rewrites. A force push destroys branch history without deleting the repo. |
| Work item recycle bin: holds items until restored or purged | No API coverage. Items deleted through the API or a script never reach the bin. |
| Test Plans and packages: 14 to 30 day windows | Shortest retention on the data assessors ask about most. |
| Pipeline runs: 30-day default, up to 730 with leases | Every native copy shares one blast radius, same tenant, same credentials as the incident. |
| Storage: Microsoft-managed replication only | No independent copy. No storage you own, no immutability you control. |
Five Failures Recycle Bins Don't Cover
- Bad automation. A script with broad permissions bulk-deletes work items.
- History rewrite. A force push destroys branch history everyone assumed Git would protect.
- Cleanup cascade. An admin deletes a project and the deletion reaches further than planned.
- Compromised token. A personal access token corrupts or removes everything it can reach.
- The audit request. An assessor asks about a release from eighteen months ago.
Key Capabilities
Nothing left to reconstruct from memory Repos, Boards, Pipelines, Test Plans, and Artifacts are protected as one system, along with wikis, dashboards, process configuration, teams, and service hooks.
One failure stays one failure Restore a repository with its branches and policies, a work item with its fields and attachments, a wiki, a team, or a whole project, in place. No standing down the delivery org to retrieve one object.
The copy outlives the incident Backups land in storage you own, in the account and region you choose: Amazon S3, Azure Blob, Google Cloud Storage, Wasabi, Dell Data Domain, or any S3-compatible target. Lock it immutable so a compromised token or a mistaken admin action can't reach it.
Protection runs without anyone watching it Incremental, policy-driven backups on your schedule. No export scripts to maintain, no one to remember to run them.
The record is still there when someone asks for it Retained work items and test plans produce the release evidence an assessor needs, months or years later.
One console for the whole Microsoft estate Azure DevOps runs under the same policies, console, and role-based controls as Microsoft 365, Entra ID, Azure, and GitHub.
At a Glance
| Full Delivery System Coverage | Repos, Boards, Pipelines, Test Plans, Artifacts, and the project and org settings that hold them together. |
| Granular Recovery | Restore a repository, work item, pipeline, wiki, or team in place, without disturbing anything still running. |
| Your Storage, Immutable | Backups land in storage you own, in your account and region, locked immutable if you choose. |
| Evidence That Survives | Retained work items and test plans produce the record an assessor asks for, on retention and residency you set. |
| One Console, Whole Estate | Azure DevOps joins Microsoft 365, Entra ID, Azure, and GitHub under one policy engine. |
| Your Backup, Your SLA | 30-minute, hourly, daily, or any custom frequency. Retain as long as you need. Restore in minutes. |
