Moving live servers and network kit is one of those jobs where small oversights become long nights. A missing patch label, an unbooked loading bay or a PDU with the wrong sockets can turn a planned four-hour window into a weekend. This data centre migration checklist is written for IT managers and MSPs who need to move equipment between racks, data halls or entire data centres, in the UK or anywhere else, and want the physical side to go as smoothly as the logical side.
This checklist concentrates on the hands-on work: what to record, what to book, what to bring and what to check before you hand the environment back to users.
Phase 1: Discovery and audit
You cannot move what you have not documented. Start with an honest audit of the current environment.
- Asset register: make, model, serial number, asset tag, rack and U position for every device, including blanking panels, shelves and KVM.
- Cable map: record every power, copper and fibre connection end to end, including port numbers on both sides. Note fibre type (single-mode or multimode) and connector type.
- Photographs: take front and rear photos of every rack, plus close-ups of dense patching. They are invaluable at 2am when a label has fallen off.
- Power profile: note each device’s PSU count, connector type (C13 or C19) and whether it is fed from A and B supplies.
- Dependencies: map which applications run where, what talks to what, and which systems must come up first (typically storage, then core network, then hypervisors, then applications).
- Health check: flag degraded disks, failed PSUs or warning lights now. Older kit that has run continuously for years can fail to restart after a power cycle, so you want to know its condition before you touch it.
If you do not have staff at the current site, a smart hands engineer can carry out the audit, photograph each rack and return a populated register for you to plan against.
Phase 2: Design the target environment
Treat the move as a chance to fix what has grown messy, but do not let redesign creep into the cutover itself.
- Rack elevations: draw the new layout U by U. Keep heavy items low.
- Power capacity: confirm the per-rack power allocation at the new site, and that PDUs have enough of the right sockets on both A and B feeds.
- Weight and floor loading: check the rack’s rated load and the facility’s floor limits, especially for storage arrays and fully populated chassis.
- Cooling and airflow: confirm hot aisle and cold aisle orientation, and that devices with reversed airflow are mounted correctly.
- Connectivity: order cross-connects, carrier circuits and any new fibre early. Lead times are set by the facility and carriers, so request them as soon as the target rack is confirmed.
- IP and DNS: decide whether addressing stays the same. If anything changes, list every hard-coded IP, firewall rule, VPN peer and monitoring check that will need updating.
Phase 3: Logistics and site access
This is where many moves come unstuck, because the details sit with facilities teams rather than IT.
- Register every engineer on the access list at both sites, with photo ID, and complete any inductions in advance.
- Book loading bays, goods lifts and parking for the move window.
- Check whether the facility needs a method statement or risk assessment, and submit it in good time.
- For moves between sites, arrange appropriate transport: anti-static packaging, padded crates and, for valuable or sensitive kit, a vehicle that is not left unattended.
- Agree chain of custody for drives and storage media, including who signs for them at each handover.
Phase 4: Pre-migration preparation
A week or two before the window, lock things down.
- Take verified backups and, where possible, test a restore. A backup you have never restored is an assumption, not a safety net.
- Freeze changes on the affected systems.
- Label both ends of every cable with a scheme that maps to your new elevation, not just the old one.
- Pre-stage the target racks: rails, PDUs, cable management and patch panels installed and tested. Good rack and stack preparation shortens the cutover considerably.
- Prepare spares: PSUs, SFPs, patch leads, rails, cage nuts and screws. Bring more than you think you need.
- Write a runbook with a named owner for each step, timings, go and no-go points and a clear rollback plan.
- Notify users, service desks and any third parties who monitor or support the systems.
Phase 5: Migration day
Power-down
- Confirm final backups have completed.
- Put monitoring into maintenance mode so alerts do not flood the team.
- Shut down in reverse dependency order: applications, then virtual machines, then hosts, then storage, then network.
- Photograph each rack again just before de-cabling.
De-rack and transport
- Remove cables carefully and bag accessories with each device.
- Use two people for anything heavy and remove PSUs or drives from heavier chassis if the manufacturer recommends it, recording which slot each came from.
- Tick each asset off the register as it leaves and as it arrives.
Install and power-up
- Rack each device in its planned position, cable to the new map and dress cables neatly.
- Power up in dependency order: network first, then storage, then hosts, then applications.
- Check every device for warning lights, failed disks and PSU redundancy before moving on.
Having engineers on site at both ends makes a real difference here. If your own team will be at the destination, a remote engineer can handle the de-rack at the source. Our guide on smart hands vs remote hands explains which level of support suits each task.
Phase 6: Testing and sign-off
Do not reopen services until the following are complete.
- Link status, speeds and duplex on every uplink, plus routing and VLAN checks.
- Storage paths and multipathing confirmed healthy.
- Application smoke tests run by people who know the systems, ideally against a pre-agreed test script.
- Out-of-band management (iLO, iDRAC or similar) reachable from the network.
- Monitoring taken out of maintenance mode and showing green.
- Pull a PSU or switch off one feed on a sample device to prove A and B redundancy actually works.
If a go or no-go point fails, follow the rollback plan rather than improvising. A delayed move is far cheaper than a prolonged outage.
Phase 7: After the move
- Update the asset register, rack elevations, cable map and network diagrams.
- Close out cross-connects and circuits at the old site, and arrange secure data destruction for any retired drives.
- Return access cards and confirm the old space has been handed back cleanly.
- Hold a short review to capture what worked and what to change next time.
Rack migration within the same building
Moving kit between racks or halls in the same facility is simpler, but the same rules apply. You still need a cable map, labelling, a power check on the new PDUs and a rollback plan. The common mistake is to treat an internal rack migration as a quick job, then discover the new rack has no spare C19 sockets or the patch panel ports do not line up.
Where field engineering support fits
Many IT managers and MSPs keep the planning in-house and bring in engineers for the physical work, particularly when sites are in different cities or countries. Simplify Networks provides vetted engineers for audits, de-racking, installs and post-move checks across the UK and worldwide, as part of our data centre services. MSPs can also use us under their own brand through our white-label field services, and you can see a related engagement in our London data centre CPU upgrade case study.
Planning a data centre migration?
Tell us what is moving and where, and we will help you put the right engineers on site at each end of the move.