At a high level, this is what the DMS architecture looks like:
Both the Source and Target databases are external to DMS. They are represented internally by endpoint resources that are references to the databases. Endpoints can be reused between different tasks if needed.
This recipe starts by defining the replication instance details. Keep in mind that the DMS migration process works best when the migration/transform between the two databases is kept in memory. This means that for larger jobs you should allocate a more powerful instance. If the process needs to temporarily write data to disk (such as swap) then the performance and throughput will be much lower. This can have flow-on effects, particularly for tasks that include ongoing replication.
Next, the two endpoints are defined. It is very important to verify your endpoint configuration by using the built-in testing feature so that your tasks do not fail later in the process. Generally, if the connectivity test fails, it is one of two main issues:
- Network connectivity issues between the replication instance and the database. This is particularly an issue for on-premise databases, which are usually specifically restricted from being accessed externally.
- User permissions issues: For example, in the case of MySQL, the root user cannot be used to connect to the database externally, so this default user cannot be used.
Defining the task involves defining your migration type. The recipe uses the simplest type; migrate tables. This means that the data will be copied between the two databases, and will be complete when the data is propagated. We also get to define the behavior on the target database. For simplicity, we have configured the task to drop the tables in the target database ensuring that the two databases look as similar as possible, even if the tables are renamed, or the table mappings change. For the task table mappings we use the wildcard symbol % to match all tables in the source database. Obviously, you could be more selective if you only wanted to match a subset of your data.
Once the replication instance, endpoints, and task are defined the wizard ends and you are returned to the DMS console. After the task is finished creating it can be started.
As it is a migrate existing data-type task, it will complete once all the data has been propagated to the target database.