AWS Migration Cutover Options


AWS Migration Cutover Options

1. Traditional Network Cutover (DNS / IP / Routing Shift)

This is the standard approach when migrating workloads using replication tools like AWS Application Migration Service (MGN) or database migration tools.

How it Works

  • Keep on-prem server running until the final synchronization.
  • At cutover time, shift traffic by:
    • Updating DNS records
    • Changing routing / firewall rules
    • Moving virtual IPs in advanced setups

Pros

  • Minimal downtime (often just minutes for DNS / TCP drain)
  • Predictable transition if planned carefully
  • Users often don’t notice

Cons

  • Requires clean network alignment (firewalls, VPN/VPC connectivity, DNS changes)
  • Higher operational complexity
  • Harder in environments with legacy networks

2. Application Shutdown Option (Instead of Network Cutover)

A simpler and more controlled approach is to intentionally shut down the on-prem application during cutover, then bring it up only in AWS.

How it Works

  1. Announce a maintenance window.
  2. Shut down the application on-prem.
  3. Finalize replication (last sync is clean, no data churn).
  4. Start the migrated instance in AWS.
  5. Run validation and open traffic to AWS version only.

Pros

  • Radically simpler cutover (no real-time switch, no race conditions)
  • Guarantees data consistency (no writes occurring during final sync)
  • Eliminates risk of split-brain between on-prem and AWS
  • Great for:
    • Stateful apps
    • Legacy apps
    • Systems with fragile network dependencies
    • Low-tolerance for replication errors

Cons

  • Requires a downtime window
  • Users may have to plan for brief outage
  • Not suitable for real-time, 24×7 systems without maintenance windows

When Shutting Down Makes More Sense

Use this method when:

  • The application cannot tolerate write divergence
  • Networking between on-prem and AWS is unstable or complex
  • DNS propagation impact is unpredictable
  • Migrating a monolith with tightly coupled components
  • Expect significant deltas during final sync
  • App already scheduled for a maintenance window

Example Cutover Procedure (Shut-Down Approach)

  1. Notify users of maintenance window
  2. Freeze application changes
  3. Stop application services on the on-prem server
  4. Stop background jobs / schedulers
  5. Perform final replication sync (AWS MGN, DMS, rsync, etc.)
  6. Power off or isolate on-prem server
  7. Launch AWS instance and test integrity
  8. Run smoke tests
  9. Update DNS to point to AWS system
  10. Open application to users

Summary

Approach Pros Cons Best For
Network Cutover Near-zero downtime, smoother for users Higher complexity, network friction Modern systems, stable networks
Shutdown, then Cutover Maximum data integrity, safest Requires downtime Legacy apps, high-risk migrations