Rack and Stack: What It Means, the Full Process, and a Checklist for Data Center Teams
What rack and stack means, the full sequence from receiving to handoff, a copyable SOP checklist, common mistakes, and a rubric for checking a technician's rack and stack skill.
TL;DR
- Rack and stack is the physical installation of IT equipment in a data center rack: mounting rails, lifting the device in, connecting power and network, labeling, and recording it.
- The job starts before the lift. Check the delivery against the order, confirm the rack elevation, power budget, and weight, and only then mount the rails.
- Most rack and stack errors show up later: a server on one power feed, an unlabeled cable, a blocked airflow path, or a record that no longer matches the rack.
- Use one written checklist for every install, and check a technician's skill by watching them work through it, not by asking whether they have done it before.
What is rack and stack?
Rack and stack (also racking and stacking) is the physical installation of IT equipment in a data center rack: mounting the rails, lifting servers, switches, and storage into their assigned rack units, and connecting, labeling, and recording them. It turns a delivered box into a device that is powered, cabled, documented, and ready for the network or systems team.
The term is plain about the two physical motions, but the work around them is what separates a good install from one that causes an outage months later. A complete rack and stack covers receiving, planning, mounting, power, cabling, labeling, documentation, and a verified handoff. It is usually one of the first tasks a new IT data center technician is trusted with, and it is a common smart hands request at colocation sites.
Outside data centers, "rack and stack" also means ranking people or priorities in order, a usage common in the military. This guide covers only the data center meaning.

The rack and stack process, step by step
A rack and stack job has ten steps, from receiving the equipment to handing it off. Each step has an output that the next step depends on, so skipping one usually means redoing work later.
| Step | What the technician does | Output |
|---|---|---|
| 1. Receiving and inspection | Matches the delivery to the order and ticket, checks for shipping damage, records serial numbers and asset tags | A confirmed, undamaged device tied to a ticket |
| 2. Rack prep and elevation | Confirms the target rack, rack units, power, weight, and airflow against the approved elevation | A clear, verified space in the rack |
| 3. Rails | Installs the rail kit at the right rack units, in the right hole type, level front to rear | Rails locked into both posts |
| 4. Lifting | Plans the lift, uses a server lift or a second person for heavy equipment | The device at rail height without strain or drops |
| 5. Mounting | Slides the device onto the rails until the locks engage and secures it | A seated, locked device |
| 6. Power | Connects each power supply to its assigned PDU and outlet, A and B feeds separate | Redundant power on the planned circuits |
| 7. Cabling and labeling | Connects network and management cables to the assigned ports and labels both ends | Correct, traceable connections |
| 8. Cable management | Routes and dresses cables without blocking airflow or the rails | A rack that can be serviced and cooled |
| 9. Documentation | Updates the DCIM or asset records, rack elevation, and port maps | Records that match the rack |
| 10. Verification and handoff | Checks lights, power on both feeds, link status, and photos, then closes the ticket | A device the next team can use |
The order matters most at the boundaries. Power goes on before data cables only if the site's procedure says so, and documentation is part of the job, not an afterthought done at the end of the shift. If you write this sequence into a site SOP, put it through your change process: Uptime Institute's guide to writing a good method of procedure notes that changes of state at a site "operate within a change management environment."
Before the lift: receiving, inspection, and the rack elevation
Most rack and stack problems can be prevented before anything is lifted. The technician should confirm that the right device arrived undamaged and that the rack has the space, power, and weight capacity the plan assumes.
Receiving and inspection
- Match the model, quantity, and serial numbers to the purchase order and the work ticket.
- Inspect the box and the device for shipping damage, bent rails or ears, and loose parts. Photograph damage before you unpack further.
- Check that the rail kit, power cords, and any optics the order includes are in the box, and that the cords have the right plug type for your PDUs.
- Unpack outside the data hall where the site requires it, to keep cardboard and packing dust off the floor.
- Record the asset tag and serial number before the device goes into the rack, where they may be hard to read.
Rack prep and the elevation plan
A rack elevation is the plan that shows which device goes in which rack units. Confirm it against the physical rack before you start: the rack units are empty, blanking panels can be removed, the rack is the type the rail kit needs, and the PDU outlets on the plan are free.
Check rack compatibility in the server's own install guide. Oracle's rack installation chapter for its Sun Fire X4170, X4270, and X4275 servers, for example, requires a four-post rack that conforms to ANSI/EIA 310-D-1992 or IEC 60927, states that two-post racks are not compatible, and gives a range of 610 to 915 mm between the front and rear mounting planes for its slide rails. Other servers have other limits, so read the guide for the model you are installing.
Rails, lifting, and mounting the equipment
Install the rails at the rack units on the elevation, using the hardware for that rack's hole type, then lift the device onto them with a lift or a second person. The device is mounted only when its rail locks engage and it cannot be pulled free.
Rails
Rail kits are either tool-less, which latch into the rack holes, or bolt-on. For bolt-on rails, Oracle's guide says to use the correct screws for threaded posts, first checking whether the threads are metric or standard, and to use cage nuts in racks without threaded holes. It also says to attach the rails without fully tightening the screws, set the rail width with the spacer, and only then tighten them.
- Count rack units from the rack's own markings, not by eye, and confirm the front and rear rails are on the same unit.
- Make sure each rail is the correct side (left or right) and the correct way round (front and rear).
- On tool-less rails, check that every pin is fully seated in the hole at both ends before loading the rail.
- Check that the rails do not foul the PDUs, cable managers, or the rear door.
Lifting and team lifts
Heavy equipment should go in with a lift or two people. In its guidance on controlling ergonomic hazards, OSHA lists "use a device to lift and reposition heavy objects" as an engineering control and "require that heavy loads are only lifted by two people" as a work practice control, and says engineering controls are the most desirable where possible. Oracle's guide states that installing those servers "requires a minimum of two people because of the weight of the server." Set your site's own lift limits with your safety team and write them into the SOP.
Mounting
Align the device's inner rails or brackets with the rack rails, slide it in until the release stops, press the release, and push it home until the locks click. Then confirm it is locked before you let go or start cabling. If the device uses a cable management arm, install it now so the cables are routed through it from the start.

Power: A and B feeds and PDU outlet assignment
Connect each power supply to the PDU and outlet named in the plan, and put a device's redundant power supplies on different feeds, usually called A and B. If both cords go to the same PDU, the device loses its redundancy, and nothing on the front panel will show it.
- Read the outlet assignment from the plan or DCIM before you plug in. Do not move to a nearby free outlet because it is closer.
- Trace each cord from the power supply to the PDU. Color-coded cords or cord labels for A and B make the check faster for the next person.
- Check that the plug type and the PDU's outlet type match, and use a locking cord or retention clip where the site requires it.
- After power-on, confirm on the PDU or monitoring system that each feed shows the new load, so you know both cords are live.
- Do not add load to a PDU or circuit that the plan does not cover. Ask the capacity owner.
Oracle's guide includes cautions on circuit overloading and on keeping reliable earthing when equipment is connected through power strips rather than directly to the branch circuit, and asks installers to consider the equipment's nameplate power ratings. The power budget is usually set by a planner, but the technician is the last person who can catch a mismatch before it is live.

Cabling, labeling, and cable management
Connect each cable to the port in the port map, label both ends, and route it so it does not block airflow, rails, or neighboring equipment. A cable that works but cannot be traced is a future outage during the next change.
Labeling to TIA-606
Most sites base their labels on the ANSI/TIA-606 standard for administering telecommunications infrastructure. Brady's overview of TIA-606-C describes it as a voluntary standard and lists racks and cabinets, patch panels, ports, cabling, and the data center room grid among the things to identify. It summarizes the label criteria introduced in TIA-606-B: legible, permanent, placed at both ends of cables, printed with the termination point, and matched with records. AKCP's summary of ANSI/TIA-606-D says that revision was published in October 2021 and supersedes 606-C, and describes identifiers that can include the floor, room, cabinet grid location, patch panel position in rack units, and port.
Your site's labeling scheme is what the technician must follow. Put an example label for each type (device, power cord, data cable, patch panel port) in the SOP so every installer prints the same format.
Cable management
- Route cables along the rack's cable managers, not across the front or rear of other devices.
- Leave enough slack for the device to slide out on its rails, or route through the cable management arm, and no more.
- Use hook-and-loop straps rather than tight plastic ties that can crush cables.
- Keep power and data separated as the site standard requires.
- Do not block the device's air intake or exhaust, and refit blanking panels in empty rack units.
- Respect the bend radius of fiber and copper; clean and inspect fiber connectors before you mate them.
Documentation, verification, and handoff
Update the records before you close the ticket, then verify the install against them. The job is complete when the rack, the DCIM or asset system, and the ticket all say the same thing.
Documentation
- Asset record: model, serial number, asset tag, owner, and the ticket number.
- Location: room, row, rack, and the rack units the device occupies.
- Power: PDU and outlet for each power supply, with the feed (A or B).
- Network: switch, patch panel, and port for each cable, matching the labels.
- Photos of the front and rear after cable management, if the site requires them.
Verification
Oracle's guide ends its rail installation with a check: pull the server out to its stops, inspect the cables for binding or kinks, and confirm that the cable management arm extends and retracts without binding. Add the checks your site needs: both power supplies healthy, link lights on the expected ports, no fault lights, the remote management interface reachable if the ticket includes it, and blanking panels in place.
Then hand off in writing. A short ticket note should say what was installed, where, on which feeds and ports, what was checked, and anything that differs from the plan. If something did not match the plan, stop and escalate before you improvise: Uptime Institute's MOP guidance says procedures should include a back-out plan for returning a system to a stable state.
Rack and stack checklist (SOP template)
Copy this rack and stack checklist into your SOP and replace the items in square brackets with your site's rules. Keep it short enough to use at the rack, and have the technician initial each section on the ticket.
1. Before you start
- Approved ticket or change [number] in hand, with the rack elevation, power plan, and port map attached.
- Delivery matched to the order; serial numbers and asset tags recorded; no shipping damage.
- Rail kit, power cords [plug type], and optics present and correct for the rack and PDUs.
- Target rack units empty; rack type and hole type match the rail kit; anti-tip bar or stabilizer in place.
- Planned PDU outlets free on both the A and B feeds; device weight within the rack's rating.
- Lift plan agreed: server lift or two people for anything over [site lift limit].
2. Install
- Rails installed at the planned rack units, correct side and direction, seated at front and rear.
- Device lifted with the lift or a second person, slid onto the rails, and locks engaged.
- Device secured with screws or latches as the vendor guide requires.
- Cable management arm installed if used.
3. Power and cabling
- PSU 1 to [PDU A, outlet] and PSU 2 to [PDU B, outlet], as planned; cords labeled.
- Data and management cables connected to the ports on the port map; fiber cleaned and inspected first.
- Both ends of every cable labeled in the site format.
- Cables routed through managers with service slack; no airflow, rail, or door obstruction; blanking panels refitted.
4. Verify, record, and hand off
- Device slides out and back without cable binding; locks re-engage.
- Both power supplies healthy; load visible on both feeds; no fault lights.
- Link lights on the expected ports; remote management reachable [if in scope].
- DCIM or asset record, rack elevation, power map, and port map updated.
- Photos taken [if required]; ticket note written with anything that differs from the plan.
- Work area clean, packaging removed, rack doors closed.
Common rack and stack mistakes
The costliest rack and stack mistakes are the ones that look fine on the day: a device on a single feed, a mislabeled cable, or a record that does not match the rack. Each of them shows up later, usually during a failure or a change.
- Both power cords on one feed. The device runs, but it goes down with that feed.
- Off-by-one rack units. Rails one unit off at the front or rear, or a device installed one unit away from the elevation, so the records are wrong from day one.
- Wrong hardware for the hole type. Rails forced into a rack they were not made for, or missing cage nuts, so the rail is not secure.
- Lifting alone. A heavy device lifted by one person to rail height, risking an injury, a dropped server, or a bent rail.
- Top-heavy loading. Heavy devices placed high in a rack that is otherwise empty, without the anti-tip bar out.
- Labels on one end only. The cable can be traced from the server but not from the switch.
- Cables that block airflow or service. Bundles across the exhaust, no slack to slide the device out, or blanking panels left on the floor.
- Records updated later. The DCIM entry is left for the end of the shift, then forgotten or done from memory.
- Improvising when the plan is wrong. Picking a nearby outlet or port instead of stopping and escalating.
How to assess a technician's rack and stack skill
Assess rack and stack skill by watching the person do it against the checklist, with criteria you can observe. Ask them to rack, cable, and record a device to a given elevation, power plan, and port map, and score what they do, not what they say they would do.
| Criterion | Not yet | Meets the standard | Strong |
|---|---|---|---|
| Preparation | Starts without checking the ticket, elevation, or delivery | Checks the ticket, delivery, rack units, and outlets before starting | Also spots a problem in the plan, such as weight or a used outlet, and raises it |
| Rails and mounting | Rails on the wrong unit or not seated; device not locked | Rails on the planned units, seated at both ends; locks confirmed | Checks clearances with PDUs and doors before loading |
| Lifting and safety | Lifts heavy equipment alone; anti-tip bar not used | Uses the lift or a second person and the anti-tip bar | Sets up the lift and the work area before asking for help |
| Power | Cords on one feed or on unplanned outlets | Each PSU on its planned A or B outlet; checks load after power-on | Labels cords and notes the feed in the ticket |
| Cabling and labels | Wrong ports or labels on one end only | Correct ports; both ends labeled in the site format | Cleans and inspects fiber unprompted; neat, serviceable routing |
| Documentation | Records left for later or incomplete | Records match the rack before the ticket closes | Notes every difference from the plan clearly |
| Stop and escalate | Improvises when the plan does not match | Stops and asks when the plan does not match | Proposes a fix and waits for approval |
Decide in advance which errors fail the check outright. Lifting a heavy device alone and both cords on one feed are good candidates. Plant one mismatch in the exercise, such as an outlet that is already in use, and see whether the person stops. Our data center technician skills assessment guide covers how to score practical checks fairly, and the data center technician job description shows where rack and stack sits among the other skills you hire for.
In hiring, use the result to support the interview, not as a validated prediction of job performance. It is one observed sample of how the person worked on that day.
How to train rack and stack
Train rack and stack in the same order as the checklist: explain it, show it, have the trainee do it under supervision, then sign them off. The physical skills need real rails and real weight, but the planning, power, and documentation steps can be practiced before the trainee touches a live rack.
- Teach the checklist and the site standards. Walk through the SOP, the labeling scheme, the A/B feed convention, and the DCIM screens the technician will update.
- Demonstrate on a non-production rack. Show each step, including the checks, and explain why each one matters.
- Supervised installs. The trainee works through the checklist on real tickets while an experienced technician watches and signs each section.
- Practice the decisions. Give scenarios with a wrong elevation, a used outlet, or a port map error, and have the trainee decide what to do. These are hard to stage on live equipment and suit a simulation or a tabletop exercise.
- Sign off with the rubric. Use the assessment table above, and record the date and who signed.
- Refresh when things change. New rack types, rail kits, PDUs, or labeling rules mean the checklist changes, so retrain on the differences.
Uptime Institute's MOP guidance says employees should have practiced procedures before carrying them out, through rehearsals, dry runs, and walk-throughs. For where rack and stack fits in a new technician's first three months, see our data center technician onboarding plan, and for courses that cover the basics, our review of data center training programs.
How Oasis helps
Oasis Simulations builds custom 3D simulations of data center equipment and procedures that run in a web browser. We build each scenario around your racks, PDUs, labeling rules, and SOPs, and review it with your team before anyone uses it.
Training teams can use these scenarios to practice the planning, power, and documentation decisions in rack and stack before live work, and hiring teams can use them to support interviews. The scenes on this site are examples, not a fixed catalog. You can try the rack and cables demo in your browser, read about our hiring and training programs, or see how we build a program. To talk it through, book a 15-minute call.
Frequently asked questions
What does rack and stack mean?
In a data center, rack and stack means physically installing IT equipment in a rack: mounting the rails, lifting the device into its rack units, connecting power and network cables, labeling everything, and updating the records. It is also called racking and stacking.
What is the difference between rack and stack and cabling?
Rack and stack usually includes the cables that connect the new device: its power cords and its network and management connections. Larger structured cabling work, such as new backbone runs, patch panels, or cross-connects between cages, is normally a separate job.
How many people does a rack and stack job need?
Light devices can be installed by one technician. Heavy servers and storage need a server lift or a second person; some vendor install guides require two people. Set your site's lift limit with your safety team and write it into the SOP.
Why do servers need A and B power feeds?
Servers with two power supplies can keep running if one feed fails, but only if the supplies are connected to different feeds. Plugging both cords into the same PDU removes that redundancy without any visible sign.
What labeling standard is used for rack and stack?
Many sites base their labels on ANSI/TIA-606, the voluntary standard for administering telecommunications infrastructure. It calls for legible, permanent labels at both ends of cables, identifiers for racks, panels, and ports, and records that match the labels. Your site's own scheme is what technicians follow.
How long does it take to learn rack and stack?
It depends on the equipment and the site's standards. Most teams teach it early, through demonstration and supervised installs, and sign a technician off when they can complete the checklist correctly without help.
Sources
- Installing the Server Into a Rack With Slide-Rails (Sun Fire X4170, X4270, and X4275 Servers Installation Guide), Oracle
- Ergonomics: Solutions to Control Hazards, Occupational Safety and Health Administration
- Cable Labeling Standards: ANSI/TIA-606-C, What's New?, Brady
- Cable Labeling Standards for Data Centers: ANSI/TIA-606-D, AKCP
- The Making of a Good Method of Procedure, Uptime Institute Journal
