Server & Hardware
Server Migration Plan: Testing, Cutover and Rollback
A server migration plan covers more than copying files into a new environment. Application dependencies, access, data consistency and the cutover decision belong together. Use this guide when moving to a rented server to document owners, acceptance criteria and rollback conditions.

A server migration plan starts with inventory
First identify the components to migrate. Application servers, databases, file shares and scheduled jobs may run on separate systems. List domain records, certificates, licenses, IP restrictions and external connections. Discovering unknown dependencies on cutover day increases downtime risk.
Server migration plan: scope review
| Inventory item | Information to be recorded |
|---|---|
| Installation | Name, version, vendor support and owner |
| Data | Volume, change frequency and consistency checks |
| Links | DNS, IP, VPN, API and access restrictions |
| Operations | Scheduled jobs, email and reporting |
| Permissions | Access owner, approver and access termination |
Do not store inventory and passwords in the same document. Grant the migration team necessary permissions through the approved access process. Recheck software versions and licensing for the target environment; a working legacy installation is not automatically supported after migration.
Separate downtime and data-loss expectations
RTO describes the target recovery time; RPO describes the acceptable data-loss interval. They are different goals. Agree them with the business team and test whether transfer and recovery methods meet them. Microsoft's explanation of RTO and RPO can help align technical planning with business expectations.
Applications that continually write data will change after the first copy. Define final synchronization, when writes stop and when they resume. Methods depend on the application and database; this guide provides neither universal commands nor a zero-downtime guarantee.
Prepare acceptance tests before migration
| Test | Evidence of success |
|---|---|
| User login | The designated test account can access the application. |
| Data consistency | Selected records and reports match. |
| Integration | Required external connections pass controlled tests. |
| Scheduled jobs | Jobs run in the correct environment as expected. |
| Backup recovery | A sample recovery makes the application and data usable. |
| Monitoring | Errors, resource use and access events can be monitored. |
Assign an owner and evidence to every test. “The page opens” is insufficient for order-processing or accounting software. Choose representative transactions without sharing real customer data. Record results with the date and environment.
Server migration plan: cutover sequence
- Preparation: check target systems and access; appoint the migration lead.
- Initial transfer: use a copying method suited to the application.
- Pre-test: verify dependencies and acceptance scenarios at the target.
- Cutover decision: owners assess results and decide to continue or postpone.
- Final synchronization: complete writes and routing changes in the planned order.
- Acceptance and monitoring: retest critical operations; document retirement of the old environment.
If DNS changes are needed, confirm record administration and access ownership in advance. DNS caching means users may not reach the new system simultaneously. Both environments accepting writes can create inconsistencies; manage this using application-appropriate methods.
When will the rollback decision be made?
Write rollback conditions before cutover. Failure of critical operations, inconsistent data or an exceeded window are examples. Set organizational thresholds and clearly name the decision-maker and notification recipients.
Redirecting to the old server is not always enough. Assess data created in the new environment separately. Rollback must cover data-loss risk, resynchronization and user communication. An untested rollback is not a ready solution.
Handover and closure after migration
Handover records should include application acceptance, access transfer, backups and open issues. Retire the old system only after acceptance and retention decisions are clear. Review temporary accounts, unnecessary network access and obsolete jobs; record an owner and completion date for each action.
If new capacity has not been selected, use the server rental service quotation inputs. For a broader provider change, read the IT provider transition and handover planand for a platform-specific scenario, see the Yandex Cloud migration guide for reference. This is platform-independent planning; verify application-specific technical scope separately.