How to migrate a VPS without downtime

In brief

A migration without a visible outage requires preparing the destination, copying data in multiple passes, testing away from public traffic and performing a short cutover with rollback. It is not a one-command operation.

Key points

Inventory dependencies and write the rollback first.
Synchronize data before final cutover.
Validate the app, DNS, certificates and monitoring.

In this guide

01Preparation
02Synchronization and testing
03Cutover and rollback
04Frequently asked questions
05Sources

Preparation

Record services, versions, ports, scheduled jobs, secrets, disks and external dependencies. Lower DNS TTL ahead of time when the cutover depends on an address change.

Verified backup
Equivalent capacity
Certificates and firewall
Window and owners

Synchronization and testing

Copy static data first with rsync or an appropriate alternative. Use replication or a consistent backup for databases. Test with a temporary hostname or local hosts entry before sending public traffic.

Cutover and rollback

Pause writes or use a brief maintenance window, run the final sync and switch the proxy or DNS. Watch errors, latency and data. Revert traffic when a predefined acceptance criterion fails.

Frequently asked questions

Can zero downtime be guaranteed?

Not for every system. Replication and load balancing can avoid a visible interruption, while stateful applications or rigid dependencies may still need a brief window.

When can the old server be shut down?

After validating data, traffic, backups, scheduled tasks and stability for an agreed period. Keep it available while rollback is still required.

Sources

rsync documentation
NGINX documentation — HTTP load balancing