DTS supports data synchronization and migration from sources. When more than two RDS for MySQL instances are attached to the source, a single DTS task can cause performance bottlenecks and stability risks. Configure a separate DTS task for each attached RDS for MySQL instance under the instance to improve performance and stability.
Prerequisites
Review Limits for using PolarDB-X as the source instance.Configuration options
| Solution | Configuration method | Summary |
| Solution 1 | Configure a separate DTS task for each RDS for MySQL instance attached to the instance. Map database and table names from each source instance to the destination. Provides better performance and stability than a single task for the instance. Important This solution supports only table-level synchronization or migration. |
Recommended. Better performance and stability than Solution 2, with no limit on attached RDS for MySQL instances under the instance. |
| Solution 2 | Use the instance directly as the source for a single DTS task. If more than two RDS for MySQL instances are attached, task stability and operability degrade significantly and may impact your business. | Performance and stability risks increase when more than two RDS for MySQL instances are attached to the source instance. |
Solution comparison
| Item | Solution 1 | Solution 2 |
| Performance | Multiple DTS tasks significantly improve write performance for large-scale data to PolarDB Distributed Edition . | A single DTS task for the source instance. Performance bottlenecks can occur under high write volumes. |
| Stability | High stability. Each RDS for MySQL instance under the instance runs its own DTS task. A single task failure does not affect others—you only need to recover the failed task. |
Moderate stability. A single DTS task using PolarDB distributed version as the source means any fault fails the entire task. |
| Ease of use | More complex. Requires multiple tasks with table name mapping from each source RDS for MySQL instance to the destination. Configuration effort scales with the number of attached instances and tables. | Simple. One DTS task for the source instance. |
| Resource usage | Consumes n DTS instances for n separate tasks. | Consumes a single DTS instance. |
DDL synchronization and incremental migration
Currently, DTS tasks with a source do not support DDL synchronization or incremental migration. If you run DDL operations on the source instance during a task, take these steps to prevent write failures to the destination:
- Release the DTS task.
- Clear the destination database.
- Reconfigure the task.
In certain scenarios with as the source, you can apply DDL changes without stopping the task. The following table lists supported scenarios and operations.
| Scenario | Operation |
| Use Solution 1 for table-level synchronization only. |
|
| Use Solution 2 when the synchronization or migration object is the entire database. |
|
| Use Solution 2 when the synchronization object is not the entire database. |
|