Atlas lets you restore data from a scheduled or on-demand Cloud Backup, including snapshots from different projects or organizations. The following sections describe restoring from a snapshot without Encryption at Rest using Customer Key Management. To restore from a snapshot using Encryption at Rest using Customer Key Management, see Restore from a Snapshot Using Encryption at Rest.
You can restore selected databases or collections from a snapshot. To learn more, see Restore from Selected Databases and Collections.
Note
You can only perform cross-organization restores through the Atlas UI.
Restore Considerations
In addition to the prerequisites, consider the following requirements and limitations when restoring from a scheduled or on-demand Cloud Backup.
- If the
DefaultRWConcernvalue on the source snapshot differs from theDefaultRWConcernvalue on the target cluster, Atlas overrides the value on the source snapshot with the value on the target cluster. If there is no value configured for theDefaultRWConcernon the target cluster, Atlas keeps the value ofDefaultRWConcernfrom the snapshot without explicit configuration. This may differ from the default value for that MongoDB version.
This feature is not available for Free clusters (formerly known as
M0).For
M10+dedicated clusters, Atlas restores MongoDB Search index definitions from a Cloud Backup snapshot. Atlas doesn't restore search index data, so themongotprocesses perform initial syncs for all restored search index definitions. If you've defined large search indexes on your cluster, you might experience delays during snapshot restorations.Note
When you restore the data from the snapshot, the MongoDB Search index definitions from the snapshot replace any existing MongoDB Search index definitions.
Sharded Clusters
If you are restoring from a sharded cluster, the source and target clusters must have the same number of shards.
The source and target clusters must use the same type of config server. Config servers can be either config shards or dedicated config servers.
Atlas can't restore a sharded cluster snapshot to a replica set, or restore a replica set snapshot to a sharded cluster.
Flex Clusters
Atlas can't restore snapshots from Flex clusters, dedicated clusters, or Cloud Manager to a Flex cluster.
A Flex cluster allows you to have only two snapshots. If you require having more snapshots, upgrade your cluster to a dedicated cluster.
Recommendations to Optimize Restore Times
To optimize performance and reduce the amount of time it takes to restore, follow these principles where applicable:
Select a target cluster that isn't global or multi-cloud.
Select a multi-region target cluster only if copies of the snapshot you plan to restore exist in every region of that cluster.
Select a target cluster that belongs to the same cloud provider region as the snapshot.
Select a cluster tier with the same storage capacity as the capacity of the original volume used by the source cluster.
If the target cluster runs on AWS with configured IOPS, select the configured IOPS to fall within the configured range.
Select a target cluster that is not configured to use NVMe storage. NVMe storage degrades restore performance.
For AWS source clusters created before March 27, 2024, click the Faster Restore button in the cluster's Backup details in Atlas UI to enable faster cross-project restores using the direct attach restore method. The target cluster must meet all conditions listed in this section to use this feature.
When you activate Faster Restore, Atlas replaces each node in the source cluster one at a time during your next maintenance window. If the maintenance window ends before all nodes are replaced, Atlas finishes replacing the current node, then waits for the next maintenance window to start replacing the next node. Secondary reads and analytic nodes are unavailable during the rolling replacement.
Faster cross-project restores are automatically enabled for AWS clusters created after March 27, 2024, all Google Cloud and Microsoft Azure clusters, and all clusters in MongoDB Atlas for Government. If this feature is enabled for your cluster, the Faster Restore button doesn't appear.
Fallback Snapshots
If a scheduled snapshot fails for any reason, Atlas attempts to repeat the snapshot process. If necessary, you may use the resulting fallback snapshot to restore the cluster. This isn't recommended: fallback snapshots use a different process from regular snapshots. They may contain inconsistent data.
Fallback snapshots are marked in the UI with a warning icon, and a warning message appears in the restore modal window if the restore uses a fallback snapshot.
Warning
Restoring your cluster from a fallback snapshot may result in inconsistent data across your cluster, and should be considered an option of last resort.
Required Access
To monitor a backup restore job until it completes, you must have Project Read Only access or higher to the specific project.
To start a restore job, you must have Project Backup Recovery Operator, Project Backup Manager, or Project Owner access to the project.
For cross-organization or cross-project restores, the required permissions apply to both the source and target projects. You must have the necessary permissions: Project Backup Manager or Project Owner in both projects to initiate or manage such restore operations.
Procedure
When you perform a cluster-level restore, Atlas deletes all existing data on the target cluster before the restore. The target cluster is unavailable for the duration of the restore. As part of the restore, Atlas also restores any indexes. This does not apply to collection- and database-level restore.