A data center relocation checklist is a time-boxed plan that gives every task a deadline, from ninety days out to the week after the move. Move day is mostly execution of decisions already made, and when something goes wrong in the window, the cause usually traces to a task that slipped at T-60 or was never assigned at all.
Checklists get a bad name because most of them are generic. The useful kind is time-boxed: each item has a date by which it has to be done, and a missed date is a signal to re-plan rather than a task to carry forward. What follows is a working structure for a single move event of moderate size. Larger or multi-phase programs stretch the front end, not the back.
T-90: What Has to Be Settled Ninety Days Out?
Ninety days out, the goal is a complete and agreed picture of the estate and the constraints around it. Everything later depends on it.
- Build the asset inventory from physical walk-downs, not just the CMDB. Record make, model, serial, asset tag, rack location, and power draw.
- Map application dependencies and group systems into move bundles that can go down together.
- Survey both sites. Confirm dock access, elevator capacity, floor loading, power, cooling, and circuit availability at the destination.
- Order anything with a long lead time: carrier circuits, new racks, PDUs, optics, and replacement hardware for equipment too old to risk moving.
- Name an owner for every application and every workstream.
The pre-move site survey is where most surprises get caught cheaply. A freight elevator rated below the weight of a populated rack is a design problem at T-90 and an emergency at T-1.
Carrier circuits deserve particular attention. They routinely take longer to deliver than anything else on the list, and a destination with racks and power but no WAN is not a destination.
T-60: How Do You Design the Move?
Sixty days out, with the inventory stable, you design the move itself.
Decide the move method for each bundle: lift and shift, swing hardware, or data migration ahead of a physical move. Draw the destination rack elevations and the cabling plan. Write the shutdown and power-on sequence for each bundle, and agree the go/no-go criteria and rollback points with the business.
This is also the point to consolidate everything into one document. A migration workbook that serves as the single source of truth prevents the slow divergence between the spreadsheet the network team uses and the one the movers were given.
Confirm insurance, transport tier, and custody requirements. If any data-bearing equipment is regulated, for example under HIPAA device and media controls, the documentation requirements need to be settled now, not discovered at the loading dock.
T-30: What Should Be Ready at Both Ends?
Thirty days out, the destination should be physically ready and the source should be under control.
- Install and burn in racks, PDUs, and structured cabling at the destination. Test every power feed under load.
- Stand up and test the destination network, including circuits, firewalls, and out-of-band access.
- Pre-stage anything that can be pre-staged: rails, cable bundles cut and labeled, patch panels populated.
- Have your IT team take and verify a restorable copy of every system in the move. A copy nobody has restored is an assumption.
- Issue the change freeze and communicate the outage window to stakeholders.
Labeling happens here too. Every device and every cable should carry an identifier that maps to the destination elevation, so the crew at the far end never has to guess.
T-7: Rehearse and Confirm
The final week is for confirmation, not new work. Anything still being built at T-7 is a risk to be logged and decided on.
Walk through the runbook with every team, in order, timing each step. Confirm crew names and site access credentials at both facilities. Reconfirm the truck, the route, and the delivery appointment. Check that vendor support contacts are available during the window and that any required hands-on vendor technicians are booked.
Run a final reconciliation of the inventory against the rack elevations. A device found in the rack but not on the list is a device that will either be left behind or arrive without a destination.
T-1 and Move Day
The day before, confirm the final go/no-go with the business, confirm with your IT team that the latest copy jobs completed, and pre-position packing materials, carts, and tools at the source.
On move day the sequence is familiar: graceful shutdown in the agreed order, de-cabling, de-racking, packing, loading with custody documentation, transport, receipt and inspection, racking, cabling, and power-on in reverse order. The work is manual and parallel, and the risk is that decisions get made in a hallway rather than in the record.
That is why the command structure matters as much as the crew. A single point of coordination, a timestamped log, and a clear owner for the go/no-go and rollback calls are what separate a controlled window from a long night. For programs without in-house capacity to run that structure, dedicated project planning and management covers the command function alongside the physical work.
Application owners test and sign off before the window closes. A system that powers on is not a system that works.
T+7: Stabilize and Close
The week after the move catches what move day missed. Drives that survived the first power-up sometimes fail in the following days. Monitoring thresholds that were suppressed during the window have to be re-enabled. Cabling that was patched temporarily needs to be finished properly.
- Re-enable all monitoring and alerting and review what fired.
- Update the CMDB and asset records with final locations.
- Close out custody documentation and file it where auditors can find it.
- Decommission or clear out the source site on its own schedule.
- Hold the post-move review while memories are fresh.
The checklist is only as good as the dates on it. Put each item on the plan with an owner and a deadline, and treat a missed deadline in the first sixty days as information about whether the move date is still realistic. Moving the date at T-60 is an inconvenience. Moving it at T-7 is expensive, and not moving it when you should have is how your data center relocation turns into a recovery exercise.