To handle the resources required for data replication, AWS DMS Serverless transitions through different stages that reflect the internal processes of the service. When a replication is initiated, AWS DMS Serverless assesses the workload, allocates the necessary resources, and begins data replication based on the following states.
Initializing: AWS DMS Serverless kicks off replication by setting up the necessary parameters for the task. Initialisation takes around 2 to 5 mins.
Preparing Metadata Resources: The system establishes a connection to your source database to gather metadata required for replication. This phase may take around 15 to 20 mins.
Testing Connection: DMS Serverless checks if connections to both source and target databases are working properly. Testing source and target connection may take upto 2 to 5 mins.
Fetching Metadata: Metadata from the source is retrieved, allowing DMS Serverless to analyze database structure and capacity needs. It may take upto 5 mins to analyze metadata.
Calculating Capacity: Based on the fetched metadata, DMS estimates the amount of resources required to handle replication tasks. This task may take 1 to 2 mins.
Provisioning Capacity: AWS DMS allocates the necessary compute resources based on the earlier calculations to support the replication. It might take anywhere between 15 to 20 mins to provision serverless replication.
Replication Starting: Data replication begins, including phases such as full load and Change Data Capture (CDC), which captures ongoing changes. It takes 1 to 2 mins to start replication process.
Running: Continuous replication is now underway, keeping source and target systems in sync. It can transition to a stopped state based on user actions.