Riverton Analytics runs Microsoft SQL Server on premises with two read-only replicas, and the nightly differential backups for 28 databases have become expensive and difficult to manage. They plan…

Question

Options

• ACreate a Compute Engine VM, install SQL Server, and import the backup file
• BCreate a Google Kubernetes Engine cluster, deploy SQL Server, and restore the backup file
• CConfigure Cloud SQL for SQL Server as a subscriber to the on premises publisher and replicate
• DTake a native SQL Server backup to Cloud Storage and restore it into a new Cloud SQL for SQL

• A
  8% (2)
• B
  17% (4)
• C
  71% (17)
• D
  4% (1)

Explanation

The key constraints are: 28 SQL Server databases, cutover under 20 minutes, minimal disruption, Google-recommended practices. Option C uses SQL Server's native transactional replication by configuring the on-premises SQL Server as the publisher and Cloud SQL for SQL Server as the subscriber. This establishes a continuous, live data sync before any cutover event. When it is time to cut over, the application is pointed at Cloud SQL while replication is still running, meaning virtually zero data loss, and the switch itself takes only seconds per database. Completing the cutover for all 28 databases well within 20 minutes is achievable. Option A (Compute Engine + SQL Server) is not a managed service and requires manual setup, patching, and HA configuration. Option B (GKE + SQL Server) is similarly unmanaged and adds Kubernetes operational overhead. Option D (backup to Cloud Storage then restore) would require restoring 28 databases from backup files, which could easily exceed 20 minutes depending on database sizes, and the cutover would capture only a point-in-time snapshot rather than live data.

No community discussion yet for this question.