Important
The feature described on this page is not currently supported on Atlas Infinite clusters during public preview. For the features supported in public preview, see Public Preview Availability.
To migrate a cluster to MongoDB Atlas with minimal downtime, you can choose between Atlas Live Migration (which hosts Mongosync for you), or run standalone Mongosync directly. Compare these methods to decide which one meets your needs.
Use Atlas Live Migration for source clusters ≤5 TB or ≤3 shards. If your source cluster exceeds these thresholds, standalone mongosync is the better option as it provides tunable parameters, pause and resume capabilities, detailed progress tracking, and direct log access for debugging transient errors.
Differences between Atlas Live Migration and Mongosync
Migration Mode | Standalone Mongosync | |
|---|---|---|
Effort required for setting up Mongosync | Low: Fully-managed in Atlas-hosted infrastructure. Live migration (pull) is currently available in these geographic regions. | High: Requires self-hosted Mongosync infrastructure setup. |
Source/Destination cluster locations | Source cluster can be either on-premises or in Atlas. Destination cluster must exist in Atlas. | Source cluster can be either on-premises or in Atlas. Destination cluster can be either on-premises or in Atlas. |
Private networking/VPC Peering support | ||
Post-migration data verification | ||
Migration of specific collections or databases | with filtered sync | |
Tunable mongosync parameters | ||
Reverse sync/migration | ||
Pausing and resuming a migration | ||
Detailed migration logs | ||
Support for a destination cluster with preexisting data | ||
Support for source clusters with no authentication enabled |