Google ACE Deploying and Implementing a Cloud Solution Practice Question

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✓

  Use Database Migration Service (DMS)

  Why this is correct

  Database Migration Service (DMS) is a fully managed service that automates the entire migration process, including a one-time full load followed by continuous change data capture (CDC) from the source database. This approach keeps the on-premises database online and synchronized during the migration, so you can cut over with minimal downtime. For a large database, DMS handles the heavy lifting of snapshotting and applying changes without needing to manually create or transfer dump files.

• ✗

  Use pg_dump to export and pg_restore to import

  Why it's wrong here

  Using pg_dump and pg_restore requires you to manually export the source schema and data into a flat file, then upload it and restore it into the destination Cloud SQL instance. You must quiesce writes to the source database during the dump to ensure a consistent snapshot, which typically means a maintenance window and unacceptable downtime for a large production database. The dump file also grows with the database size, making the process slow and dependent on network transfer.

• ✗

  Use gcloud sql import to import a dump file

  Why it's wrong here

  The gcloud sql import command can only import dump files that already exist in Cloud Storage, so you must first create the dump, move it to a bucket, and then run the import command. This still requires the same manual dump and transfer steps as pg_dump, and you must manage the Cloud Storage bucket and permissions yourself. Additionally, the import is a one-time bulk load with no ability to capture ongoing changes, so the source database must be offline or write-enabled only after import completes.

• ✗

  Set up a Compute Engine instance to run pg_dump and then gcloud sql import

  Why it's wrong here

  Setting up a Compute Engine instance just to run pg_dump adds unnecessary infrastructure: you must provision and configure the VM, install PostgreSQL client tools, run the dump, and then copy that dump to Cloud Storage before invoking gcloud sql import. Each of these steps introduces potential misconfigurations, such as insufficient disk space on the VM or network egress costs, and you are still left with a manual, one-time snapshot that requires downtime. This option effectively combines the drawbacks of the previous manual approaches while adding more moving parts.

Go deeper

Related to this question