Common reasons for a data centre exit
- Co-location contract renewal. Prices for power and space have risen sharply, and many contracts lock you in for three to five years.
- Office move or consolidation. Nobody wants to build a new server room.
- Hardware end of life. The storage array, hosts and network kit are all due for replacement at once.
- Resilience. A single server room is a single point of failure for fire, flood and power.
How we deliver a data centre exit
Phase 1: Discovery and triage (weeks 1–6)
We inventory every server, appliance and network dependency using Azure Migrate's discovery and dependency analysis, alongside interviews with application owners. Each workload is triaged:
| Decision | What it means | Typical share |
|---|---|---|
| Rehost | Move as-is to Azure VMs | 50–70% |
| Replatform | Move to a managed service (e.g. Azure SQL MI, Azure Files) | 10–25% |
| Replace | Switch to a SaaS product | 5–15% |
| Retire | Switch off, as no longer needed | 10–20% |
Phase 2: Landing zone and connectivity (weeks 4–10)
Build the Azure foundation (identity, networking, security policy, backup, monitoring) and connect it to your sites over VPN or ExpressRoute.
Phase 3: Migration waves
Low-risk workloads move first to prove the process, followed by core business systems. Each wave includes testing, a planned cut-over and a rollback window.
Phase 4: Decommission and hand back
Data is securely wiped, hardware is disposed of or recycled with certificates, and racks are cleared so you can hand back the space on time.
Keeping the exit on schedule
- Fix the date, flex the approach. If time runs short, rehost now and modernise later.
- Weekly programme reporting against the exit date.
- Early vendor engagement for any third-party application that needs re-licensing or vendor support in Azure.
Planning a migration? Get a free assessment covering scope, ballpark cost and funding eligibility.
Get my free assessmentFrequently asked questions
How far in advance should we plan a data centre exit?
Start at least 9–12 months before your contract end or move date for estates over 100 servers. Smaller server rooms (under 30 servers) can often be exited in 3–4 months.
What happens to servers that can't move to Azure?
Some systems, such as specialist hardware, telephony or legacy appliances, can't move directly. Options include replacing them with SaaS, small edge hardware or Azure Local at a remaining site, or short-term hosting. We identify these early so they don't hold up the exit.
How do our offices connect to Azure after the exit?
Usually through a site-to-site VPN for smaller organisations, or ExpressRoute (a private connection) where you need guaranteed bandwidth and latency. Users working remotely connect through the same secure network design.
Can we exit in stages rather than all at once?
Yes, and we recommend it. Workloads move in waves over weeks or months, with the riskiest systems last. That gives you a tested process before the critical systems move.