We migrated from AWS to GCP with minimal downtime

Description and use

Migrated PostgreSQL and ML workloads from AWS to GCP with minimal downtime.

Problem

The team needed to move their PostgreSQL and ML workloads from AWS to GCP while keeping downtime to a minimum.

Outcome

The migration completed with less than 30 minutes of downtime, improved scalability, and reduced operational overhead.

Workflow

• 1. Set up a GCP project and enable Cloud SQL and Compute Engine APIs.
• 2. Create a Cloud SQL instance with PostgreSQL and install pgvector and PostgreSQL ML extensions.
• 3. Export the AWS database using pg_dump and import into Cloud SQL.
• 4. Configure network connectivity and firewall rules between AWS and GCP.
• 5. Update application connection strings to point to the new Cloud SQL instance.
• 6. Perform a test migration and validate data integrity and application functionality.
• 7. Switch traffic to GCP during a low-usage window, monitoring performance and logs.
• 8. Decommission AWS resources after successful cutover.

Evidence

We migrated from AWS to GCP with minimal downtime

Source