Migrating a hospital system to cloud with 40 minutes of downtime

Sanjeevani ran its hospital information system, PACS imaging and administrative applications on ageing on-premise hardware across three sites. Two servers were past end of support, and a disk failure the previous year had caused a 14-hour outage across one hospital.

Any migration had to work around the fact that clinical systems cannot simply be taken offline.

We assessed 23 applications, mapped their dependencies and grouped them into five migration waves ordered by clinical criticality, leaving the most sensitive systems until the process was proven.

Databases moved using continuous replication so the cutover window was a redirect rather than a restore. Every wave had a tested rollback that could be executed inside 15 minutes.

Health data made compliance a first-class constraint rather than a review item. Encryption at rest and in transit, private endpoints, role-based access with break-glass procedures and full audit logging were designed in from the landing zone onward.

The imaging system was the hardest piece: 14TB of historical studies with strict retrieval time requirements. We seeded the bulk over four weeks using offline transfer, then replicated only the delta at cutover.

All 23 applications migrated with 40 minutes of total planned downtime, taken across two scheduled windows at night. No rollback was required.

Infrastructure cost fell 31 percent against the previous on-premise total cost of ownership once right-sizing and reserved capacity were applied. The group now has a tested recovery objective of 15 minutes, against no meaningful capability beforehand.