Key takeaways

  • Give the engineer a confirmed list of what is leaving, with rack positions.
  • Shut systems down and deal with the data first. The engineer works from your confirmation.
  • Agree where the equipment is going and who receives it.
  • Decide what happens to cabling, rails and fixings, so the rack can be called clear.

A decommission is mostly planning with a short, physical burst at the end. The engineer can unrack a server in minutes, but only if they know it is the right one, that it is safe to power off, and where it goes next. Time lost on the day is almost always time lost on one of those questions. This checklist is what to have ready. If you are still working out whether your project is rack-level or something larger, start with IT versus data centre decommissioning.

1. The equipment list

The single most useful document is a list of what is leaving.

  • What: make, model and type of each item.
  • Where: facility, cabinet and rack position (U number).
  • Identity: serial numbers or asset tags, where you have them.
  • Anything staying: note equipment in the same rack that must not be touched.

A rack that is partly in use is common. The list is what stops a live system being disconnected by mistake, so make it explicit.

2. Shutdown and data

This stays with you, and it must happen before the engineer arrives.

  • Services stopped and dependencies checked.
  • Data migrated or dealt with according to your own policy.
  • Systems powered down, with confirmation passed to the engineer in writing.
Who handles data destruction

Data destruction is handled through a partner, not by the engineer removing the equipment. Decide in advance how it will be handled, so the equipment can be passed on once it is out. See what to expect when handing equipment to a partner.

3. Access and the site

  • Who raises the access request, and what the facility needs. Expect it to ask for the engineer’s name, employer and identification.
  • The working window. Attendance is scheduled around location, scope and engineer availability.
  • Site rules. Loading bay slots, restrictions on packaging or tools, and any limits on photography.
  • Space. Where equipment can be staged, and where boxes can be kept before use.

The loading bay was the limiting factor in our 300-server colocation project. Ask early what the site allows.

4. Packing and transport

  • Boxes or crates. Who supplies them, how many, and where they are stored.
  • Manifest. How items will be recorded against each box, for example by serial number.
  • Haulier or collection. The booked slot, the destination and who receives.
  • Fragile or heavy items. Anything needing extra hands or special handling.

5. Cabling, rails and fixings

A rack is not clear until this is settled.

  • Are patch cables and power cables going or staying?
  • Are rails and cable management going with the equipment?
  • Are there cross connects or circuits to be cancelled with the facility? That is a decision and a process for you and the facility.
  • What condition should the cabinet be in at handback?

6. Handover and record

Agree what you want back after the visit:

  • A list of what was removed, by serial number.
  • Any exceptions, such as equipment that did not match the list.
  • Confirmation that the racks are clear.

A final check

Before the visit, read your list as if you were the engineer. If a position, serial number or destination is missing, the engineer will have to ask on the day. For the service itself, see IT decommissioning, or tell us what is in the racks.