Rack and Stack at Scale: Building Out a New Facility

Rack and stack is the physical installation of servers, storage and network equipment into racks at a new facility: receiving, rail and device installation, power and data cabling, labeling and documentation. At scale, success depends less on the racking itself than on facility readiness, a device-level rack elevation, and controlled receiving before the first rail goes in.

Rack and stack looks like the straightforward part of a build. In practice it is where facility readiness, delivery schedules, and design decisions all converge, which makes it the point where upstream problems become visible as schedule loss.

What must be ready before equipment arrives?

The room. Racking into a room that has not been commissioned produces rework, so confirm each gate before equipment is scheduled to arrive.

The most expensive sequencing error in a buildout is accepting equipment delivery against a facility date that later slips. The equipment then needs storage, insurance, and a second handling cycle, and stored equipment has a way of arriving at the floor with damage nobody can attribute.

What should the rack elevation specify?

Everything, down to the device. The rack elevation is the controlling document, and it should be complete before anything ships.

• U-position assigned for every device, in every rack
• Weight distribution planned with heavy equipment low, and total rack weight checked against floor loading limits

• Power draw calculated per rack, with circuit assignments, A/B feed mapping, and outlet-level PDU allocation
• Phase balance planned across the rack rather than corrected afterward

• Thermal design consistent with hot and cold aisle layout, including containment and blanking
• Cable pathway capacity checked against the volume of cable it will actually carry

• Service clearance confirmed: rear access, rail travel, and cable slack for sliding equipment forward

Phase balance and pathway capacity are the two most commonly deferred items, and both are considerably harder to fix once racks are populated.

How should receiving work during a buildout?

As a control point, not a formality. Equipment arriving faster than it can be installed creates congestion that slows everything behind it.

• Schedule deliveries against installation capacity, not against vendor convenience
• Confirm dock access, appointment windows, and any lift or elevator constraints

• Size and secure a staging area, with a plan for the volume of packaging waste generated
• Inspect at receipt for external damage, and photograph anything questionable before the carrier departs

• Verify counts and capture serial numbers against the manifest at the point of receipt

Damage claims depend on identifying and documenting damage before the carrier is released. Once a delivery is signed for clean, recourse narrows sharply.

Capture Serials Once, Correctly

Scan serial numbers into the asset record at receipt, then never re-key them. Every downstream artifact (the elevation, the asset register, the warranty record, the audit package) inherits whatever accuracy is established here. Serial capture performed by transcription during installation, under time pressure, is the origin of most reconciliation problems discovered months later.

How is the physical installation done?

To the elevation, with labeling and blanking done as you go rather than afterward.

• Install rails to the correct depth, with cage nuts placed to the elevation rather than by eye
• Use lift assists and two-person handling for heavy chassis; a dropped chassis is both an injury risk and an expensive one

• Fit blanking panels as installation proceeds rather than as a cleanup pass, because the cleanup pass rarely happens
• Bond and ground racks per the design and the National Electrical Code (NFPA 70)

• Label every device and every cable at installation, both ends, machine printed (see asset tagging and labeling standards)
• Leave documented service loops sized deliberately, not coiled to fit

Labeling during installation is the highest-value habit in a buildout. Labeling afterward means tracing, and tracing a populated rack costs far more than labeling an empty one. This is the core of a professional rack and stack crew's work.

How do you make sure redundant power is really redundant?

Verify it at the outlet. Redundant power is the requirement most often defeated by ordinary carelessness during installation.

• Confirm A and B feeds originate from genuinely separate PDUs on separate circuits and separate upstream distribution
• Verify at the outlet rather than trusting the labeling on the strip

• Check that dual-corded equipment is actually plugged into both feeds
• Measure real draw after energizing and compare against the design calculation

• Verify phase balance under load and correct before the room goes into service

A rack where both power supplies trace back to the same circuit looks identical to a correctly built one until the day that circuit fails.

How do you avoid cabling delays?

Pre-measure runs and order lengths accordingly. Patch cords that are all slightly too short are the single most common cause of a re-cabling delay, and the problem is discovered only after the batch has been delivered.

Route through the designed pathways, maintain separation between power and data where required, avoid blocking airflow at the rear of equipment, and preserve access for equipment that slides. Document as installed rather than as designed. The two diverge during every buildout, and the as-installed record is the one that matters. For the backbone and cross-connect side, see structured cabling and cross-connects for a new build.

How do you commission and hand over the build?

Prove it before it carries workload, then hand over a package someone else can run from.

• Thermal survey under load, not at idle, with inlet temperatures verified across the rack face against the ASHRAE TC 9.9 envelope the room was designed to
• Power draw and phase balance measured per circuit

• Redundancy tested by pulling one feed and confirming the load carries
• Network links verified for link lights and correct patching, with speed and duplex confirmed by your network team

• Cable certification results recorded for structured runs
• Elevations, port maps, and asset records reconciled against the physical build

Then hand over the package: as-built elevations, port maps, circuit and outlet assignments, serial-reconciled asset records, and the commissioning results. A buildout is finished when someone who was not there can operate and troubleshoot it from the documentation, not when the last device is powered on.