Active Directory forest recovery From one wrong attribute to a whole forest. Backups that stay under your control.
Undo a change to one object, bring back a deleted branch, restore a Group Policy, or rebuild an entire Active Directory forest from backups taken by agents on your own domain controllers, following a plan you wrote and tested before you needed it.
- Attribute, object and batch rollback in Active Directory, Entra ID and Okta
- Forest backup sets from agents on your domain controllers, held on storage you control
- Recovery plans with a printable runbook that can be reviewed by the team before a forest restore begins
Three levels of recovery, one console
Most days you need the first. Identity CORE makes sure the third is ready too.
Object and Attribute Rollback
A change to one object, or to a hundred, undone from the saved copy taken before it.
- Active Directory, Entra ID and Okta
- Just the attribute that changed, or the whole object
- Deleted objects restored, a whole branch at a time
- Batch rollback for Active Directory
Group Policy
The policies that configure every computer, backed up as they change.
- GPMC-compatible backups on your storage server
- A readable settings report per policy
- Changes matched to the backups taken
Forest Recovery
Every domain controller backed up as a set, and a plan to bring the forest back.
- Backup sets taken at the same moment, with retention you choose
- Daily verification that every backup is present and unchanged
- Recovery plans and runbooks
- Single domain controller, staged and partition recovery
Put an object back the way it was, down to one attribute
Every collection keeps a copy of each changed object, so any object with a Backup Time can be returned to an earlier point. Find it in the Identity browser, or start from the change in the Auditor, select one object or many, and choose Rollback.
The wizard lets you revert just the attribute that changed instead of the whole object, shows every object it will touch, and asks you to pick the saved copy to return to. The Confirm step lets you verify the changes before you make them.
| Step | What you do |
|---|---|
| Objects | Check the list of objects this will apply to. |
| Attributes | Choose which details to bring back. Restore just the one thing that changed instead of the whole object. |
| Backups | Choose the saved copy to go back to. This decides what date the objects return to. |
| Options | A few settings controlling how the rollback behaves. |
| Credentials | An account with permission to write the change back to the directory. |
| Confirm | A summary of everything above for you to verify. |
An Entra ID object is reconciled against the live tenant from its saved copy, and a soft-deleted object is restored. An Okta object is restored through the Okta API. The same wizard, with the steps each directory needs.
The Rollback Preview what-if report lists the attributes that would change, with their current and backed-up values, before you commit.
Every rollback is a job in the history and an entry in System Activity: who ran it, which objects, which backup, and the result.
Group Policy backed up with every collection
A domain configured to collect Group Policy backs up the content of each changed policy from SYSVOL in the same format the Group Policy Management Console uses, so a backup can be restored with the console's own tools as well as from Identity CORE.
The settings of each policy are shown as a readable report, the Group Policy Changes report matches changes to the policy containers and links with the backups taken, and the backups are kept on your storage server for the retention you set.
Could you recover if you lost the directory? This screen answers that.
Rolling back one user is one thing. Recovering after ransomware has encrypted every domain controller is another. Forest Defender shows, forest by forest, how many domain controllers are protected, how many have actually produced a backup, and when each was last backed up.
An agent on each domain controller takes the backup and uploads it to a storage server you run. A forest job asks every agent at the same moment, so a backup set is a consistent picture of the forest, and the job warns when a set would leave out a global catalog, the DNS zones, a domain or an application partition.
-
Your storage, your keys Backups are uploaded to a storage server you install, optionally encrypted with a password that is never escrowed. For an offsite copy, the storage server can keep one in your own Amazon S3 bucket or Azure Blob Storage container. Nothing goes to a cloud you do not run.
-
Retention that keeps the last good set Sets older than the retention period are purged, but a set still in progress and the most recent completed set are never removed, whatever their age.
-
Verified daily Each storage server checks every backup it holds once a day: present, the recorded size, and the recorded hash. Missing, size mismatch and checksum failures show on the set and in the forest job's notification.
-
Agents do backups, audit events, or both The same agent can push the domain controller's audit events to the storage server, so a change to a watched object is judged within seconds.
Decide how a forest comes back before you have to
A recovery plan names a forest and the backup set to recover it from: the most recent completed set at the time, or one you pin. For every domain controller it records what the recovery does with it and the address it comes back on, so the decisions are made calmly, not at three in the morning.
Running a plan changes nothing. It works the recovery out against the backups that exist right now: the order the domain controllers are taken in, which backup each is restored from and where that backup is, the operations master roles to seize, the trusts to check, and whether the set covers the whole forest.
Put the plan on a schedule and it becomes a standing check that the forest is still recoverable as planned. A forest quietly becoming unrecoverable shows up in the notification before it matters.
- Restore, Repromote or Exclude per domain controller, with sensible defaults from the set
- Target addresses per domain controller, imported and exported as CSV for a different subnet or a test network
- A review step that refuses a plan with a restorable domain left unrestored, and warns about domains the recovery drops
- A printable runbook with the order, each backup's whereabouts, the roles to seize, the trusts, and the after-recovery checks
Start it and watch step by step
A forest recovery is the most consequential action in the product, so it needs a detailed plan.
Then the agents do the work in the plan's order: the forest root first, beginning with the schema and domain naming masters, then one domain controller per other domain, then the ones promoted again against the restored forest.
- 1
Start the recovery Give the DSRM and local Administrator passwords, or leave them blank to have a 20-character one generated per domain controller.
- 2
Restores run in order The agent on each rebuilt server fetches its backup from the storage server, verifies it against the recorded hashes, and restores the domain controller through its phases and reboots.
- 3
The rest are promoted or excluded Domain controllers without a backup are promoted again against the restored forest; the ones left out are cleaned out of it. Roles held by domain controllers that were not restored are seized.
- 4
Everything is recorded Starting, cancelling, retrying, skipping and completing a step are all audited, and the Forest Backups And Recoveries report lists the sets and the recoveries run.
Restore one domain controller from a backup set, from the agent's menu on the Forests page. Only a set taken after the forest's last recovery can be used.
Within 90 days of a forest recovery, bring back the domain controllers that were left out: Identity CORE compares the set's topology with the forest and offers the candidates to promote again.
Bring one partition of the live forest back to a backup's version, with a shadow copy taken first and the live database put back if the directory does not come up.
Common Questions
Where do the backups go?
To a storage server you install on your own network, on storage you choose. If you want an offsite copy, a storage server can also keep a copy of everything it stores in your own Amazon S3 bucket, your own Azure Blob Storage container, or both, and a backup whose local copy is gone is read back from the cloud when a restore asks for it. Backups can be encrypted with a password only you hold. Identity CORE never uploads a backup anywhere you have not chosen.
What is in a domain controller backup?
A system state backup of the domain controller taken by the agent, with a description of the forest as it was at that moment: the domains, domain controllers, their roles and addresses, the trusts and the partitions, so a recovery can be worked out from the set alone.
Does running a recovery plan touch my domain controllers?
No. A plan run only works out how the forest would be recovered from the backups that exist and writes the runbook. A recovery is a separate action that should be reviewed by the team.
Can I restore an object without the forest backups?
Yes. Object and attribute rollback uses the copies taken by the ordinary collection, so it works from the first run, with no agents and no storage server. Forest backups are only needed for recovering domain controllers.
See a forest come back in a lab
Book a demo and we will walk through a recovery plan, its runbook, and a recovery end to end.