
Perimeter, suite, cage and rack-layer credentials, escorted visitor and vendor rules, strong audit records, and hardware that stays authoritative offline.
Data centers measure security in layers: perimeter, lobby man-trap, data hall, customer cage, cabinet and rack. A data center access solution expresses each layer as a credential boundary, so an operator technician moves through shared halls, a customer engineer opens only their leased cage and cabinet, and a visiting auditor follows an escorted, time-bounded visit. Because uptime cannot depend on a network link, controllers at the critical layers hold schedules and records locally and keep operating through outages. Entries, cabinet events and override actions are retained for the audit and compliance reviews that colocation and enterprise customers run. The data centers guide covers the operating model, the offline solution covers resilience, and the government buildings page covers similar evidence-led procurement.
| Layered boundaries | Perimeter, man-trap, hall, cage and cabinet each carry a credential boundary rather than one building-wide badge |
| Customer cages | Colocation customers open only leased cages and cabinets, isolated from neighbouring tenants |
| Escorted visitors | Auditors and vendors receive time-bounded, escort-linked credentials for the visit scope |
| Audit records | Entry and cabinet events are retained for customer compliance and incident review |
| Offline authority | Critical-layer controllers enforce schedules locally so outages never open cages or block responders |
| Emergency access | Break-glass and life-safety entry is a tiny named, fully recorded procedure |
| BOM shape | Man-trap and hall controls, cage and cabinet locks, offline controllers and facility spares |
Colocation and regulated workloads require layered credentials; small private server rooms may combine one controlled suite with cabinet locks.
| Dimension | Building-wide badge access | Layered cage and cabinet control |
|---|---|---|
| Tenant isolation | A badge can reach shared halls broadly | Customers open only leased cages and cabinets |
| Visitors | Vendors move with loose sign-in | Time-bounded escort-linked visit credentials |
| Audit evidence | Perimeter log only | Hall, cage and cabinet events retained for review |
| Network outage | Controllers fail uncertain | Local schedules keep cages controlled through outages |
| Emergency | Physical key override unrecorded | Named recorded break-glass procedure with review |
| Typical fit | Single private server rooms | Colocation halls and enterprise data centers |
The man-trap and data-hall doors control people flow; cage and cabinet locks carry the tenant isolation and audit depth customers audit.
| Opening | Typical door | Lock approach |
|---|---|---|
| Perimeter / lobby | Site gate and security-reception entrance | Identity verification and visit authorization before entry |
| Man-trap | Two-door controlled lobby sequence | Anti-tailgating entry with escort and time logic |
| Data hall | Shared hall entrance doors | Operations and customer groups separated at the hall layer |
| Customer cage | Mesh cage and suite doors | Leased-cage credentials isolated between colocation tenants |
| Cabinet / rack | Individual cabinet handles and racks | Customer cabinet credentials with per-cabinet event records |
| Plant / MEP | Electrical, cooling and battery rooms | Engineering roles and vendor windows separate from data halls |
This example illustrates a typical configuration, not a specific customer project. Cage count, cabinet density and tier drive the list.
| Component | Typical logic | Notes |
|---|---|---|
| Perimeter and man-trap | Lobby identity and two-door sequence controls | Anti-tailgating logic with escort enforcement |
| Hall and cage controls | Data-hall and leased-cage readers | Tenant groups isolated between neighbouring cages |
| Cabinet locks | Per-cabinet credentials and event records | Customer-managed access within the leased footprint |
| Offline resilience | Local-schedule controllers at critical layers | Uptime preserved through network and power interruptions |
| Facility spares | Readers, cabinet locks and controllers held onsite | Swap restricted to named operations staff under the current MOQ policy |
Tenant isolation is enforced at the cage and cabinet layers rather than the building: a colocation customer’s engineer authenticates into their own leased cage and cabinets, and the same credential is rejected at a neighbouring tenant’s mesh door even though everyone passed the same lobby and data-hall entrance. Customer groups are managed per lease and reviewed when footprints change, while shared-hall access remains an operations-controlled layer. Visitor and vendor access for a customer is tied to that customer’s authorization and an escort, so a vendor booked by one tenant cannot be presented at another cage. This isolation is among the first items prospective colocation customers test during site audits.
Every non-badge-holder enters against an approved visit: the credential is time-bounded to the appointment, scoped to the halls and cages authorized, and linked to the named escort, with arrival and departure recorded. Auditors may additionally receive read-only evidence sessions of entry history for the scope they are reviewing. Vendors delivering or maintaining equipment receive cage- and time-bounded access tied to a work order, and unattended movement outside the authorized scope triggers a review. The visit workflow is tested in the pilot with a real compliance auditor, because evidence expectations vary between enterprise customers and regulated industries.
The critical layers — man-trap, hall, cage and cabinet — are specified to authenticate against locally held schedules and buffer events until the link returns, so an outage neither opens cages to anyone nor locks authorized responders out of a live incident. Controllers and readers are included in the managed-power design for the hall, and free egress and life-safety behaviour remain intact regardless of system state. Operations run a documented resilience test that pulls the network during a controlled cage entry and confirms access decisions and buffered records behave as designed. Selecting hardware on this offline behaviour matters far more in a data center than in a conventional office.
Emergency entry that bypasses normal credentials is limited to a tiny named operations and life-safety role, physically and procedurally guarded, and recorded in full with a mandatory post-event review, so the existence of an override never becomes an unlogged back door into customer cages. Fire and life-safety release follows the authority-approved design and always preserves evacuation. Customers whose contracts require notification of cage entry receive it through the documented incident process. The break-glass procedure is rehearsed in the pilot along with a simulated life-safety event, and the hardware mounting ensures the override is not discoverable or operable by visitors.
The BOM covers perimeter and man-trap controls, data-hall and cage readers, cabinet locks, offline-capable controllers, plant-room separation and onsite spares, with MOQ and lead time confirmed in writing per quotation. Operators usually pilot one hall segment and a cage row, including an escorted audit and a resilience test, before facility rollout. Send the floor plan, cage and cabinet counts, lease model and tier requirements, and the quote maps each line to an access layer rather than counting generic locks.
Send the floor plan, cage and cabinet counts and tier requirements — receive a layer-mapped BOM with written MOQ and lead-time terms.
Discuss your project (RFQ)Project configurationsLast updated: October 31, 2026 · Reviewed by the Smart Lock Factory editorial team