Migrating an ageing on-premise estate to the cloud

The situation

What was going wrong

• 01

  Core practice management ran on hardware past end-of-support with no supplier cover.

• 02

  Disaster recovery consisted of nightly tapes that had never been restored end to end.

• 03

  Deploying a change to the practice system required an evening maintenance window.

• 04

  Capacity for new offices meant buying and racking hardware months in advance.

Our approach

What we did, and why

1. 01

   Costed the target state before committing

   We modelled steady-state cloud running cost against current hosting, hardware refresh, and support spend, so the business case was based on figures rather than optimism.

2. 02

   Built a landing zone with guardrails

   Networking, identity, logging, and cost controls were established as code before a single workload moved, which prevented the sprawl that makes cloud bills unpredictable.

3. 03

   Migrated in four waves, lowest risk first

   Each wave had a defined rollback and ran in parallel until verified. The practice management system moved last, once the team had rehearsed on less critical workloads.

4. 04

   Proved recovery instead of documenting it

   We ran a full restore into an isolated environment and timed it. The first attempt exposed two gaps in the backup scope, which is precisely why the rehearsal happened before go-live.

The outcome

What changed

All workloads run on supported, patched infrastructure defined in Terraform. Verified recovery time is under four hours against a previously unknown figure. Routine deployments now happen during the working day, and the fifth office opened without any hardware procurement.

4 waves

Zero unplanned downtime

<4 hrs

Tested recovery time

−29%

Three-year cost of ownership