A new colocation deployment lives or dies on the hour you actually stand in front of the cabinet. By then the shipment is either on the dock or lost, the rails either fit or they do not, and the power either matches your gear or it does not. A short, disciplined checklist turns that moment from a scramble into a routine. This guide walks the full arc of a rack and stack, from the prep on your own dock to the documentation you hand back to your team, so nothing that could have been caught in advance surprises you on the floor. Whether your own staff handles the install or you hand it to independent rack and stack hands, the same checklist applies.
Before the gear ships
The cheapest fixes happen before a box leaves your building. Everything you confirm now is a truck roll you do not pay for later.
Confirm the cabinet, power, and access
- Cabinet or cage location, and how many rack units you actually have free. Measure, do not assume.
- Power: circuit type, voltage, phase, and receptacle. A server built for 208V single-phase will not come up on a whip you assumed was there.
- Inbound shipping: the facility receiving hours, dock address, and the exact label format they want on the pallet. Many sites reject freight that is not pre-authorized under a ticket.
- Access: who is badged, and whether an escort is required. At interconnection-heavy sites like Equinix NY2, the escort and ticketing process can gate your install window.
Label and inventory before it leaves your dock
- Serial numbers, asset tags, and MAC addresses recorded in a sheet before the lids close. Reading a serial off a rail-mounted server at the back of a hot aisle is miserable.
- Each chassis marked with its intended rack-unit position, so the install is "put box A at U20," not a puzzle.
- Rail kits, cable bundles, PDUs, and spares boxed with the matching server, not scattered across the pallet.
Pack for freight, not a car ride
- Palletize and strap. Anything that can shift will shift on a lift gate.
- Ship rails and heavy accessories separately from thin-walled chassis.
- Include a printed packing list and the deployment ticket number inside the box.
Receiving and staging on site
When the pallet lands, reconcile it against the packing list before anything gets racked. Count the boxes, check for shipping damage, and photograph the pallet as received. Stage gear near the cabinet in install order so the aisle stays clear. A good rack and stack is quiet and sequential, not a pile of open boxes blocking a hot aisle.
Rails and mounting
- Dry-fit one rail kit before committing the whole stack. Cage-nut spacing and rail depth vary by cabinet.
- Install cage nuts for every mounting point you will use, top to bottom, before lifting a single server.
- Heaviest gear low, lightest high. Keep the center of gravity down and the cabinet stable.
- Leave the planned U gaps for airflow and future growth where the elevation calls for them.
- Seat rails fully. A half-seated rail is how a server ends up on the floor during a later swap.
Power
- Land each power supply on a separate PDU or feed so a single whip or PDU failure does not drop the box. Two redundant supplies on the same strip are not redundant.
- Confirm draw against the circuit budget before energizing. Do not discover you have oversubscribed a 20A circuit at bring-up.
- Dress power cords to one side and data to the other, so a future hand can trace a run without unplugging live gear.
- Use locking C13/C14 or C19/C20 cords where the PDU supports them.
Cabling and cross-connects
- Run to length. Pre-measured or on-site-terminated patch keeps the cabinet clean; a nest of service loops traps heat and hides faults.
- Separate copper and fiber paths, respect the bend radius on fiber, and use hook-and-loop wrap rather than zip ties on structured cabling.
- Order any cross-connects early. The facility, not you, provisions the run between your cabinet and the meet-me room, and that can take days.
- Label both ends of every run as you land it, not afterward.
Labeling and the rack elevation
Every device, every port, and every cable should carry a label that matches your source of truth. Update the rack elevation (the U-by-U map of the cabinet) the moment the last device is seated. At multi-hall campuses like QTS, a cabinet with no elevation is a cabinet nobody can safely work on remotely.
Bring-up and verification
- Power on in order and watch for POST, link lights, and clean fan spin-up.
- Confirm remote management (iDRAC, iLO, BMC, or serial console) is reachable before anyone leaves the floor. This is the single most important check: if you cannot reach the box remotely, you have bought yourself another truck roll.
- Verify network reachability, and that each interface is on the port you documented.
- Photograph the finished cabinet, front and back.
Documentation and handoff
Close the deployment with a record your team can act on months later: the final rack elevation, port map, power assignments, serials and asset tags, cross-connect IDs, and the front and back photos. Good documentation is what lets someone dispatch a fix later without re-discovering the whole cabinet first. It is also the difference between a hands visit that takes twenty minutes and one that takes two hours.
Related guides
Remote hands vs smart hands
What each term really means, and which one your task actually needs.
GuideWhat is a cross-connect?
How the physical link between your cabinet and the meet-me room gets built.
GuideWhat to expect from a remote hands visit
How a dispatch runs, from the ticket to the photos you get back.
HubAll resources
Browse every guide on remote hands, colo, and data center work.
Need hands for your install?
Planning a colo deployment in the NYC metro? RemoteHands.nyc can receive your shipment, rack and stack it to your elevation, and hand back full photo documentation, day or night. Request service or call (707) 733-3342 and tell us the site and the scope.