Techbanq

Industrial security · data centres

The network that cools your data hall deserves its own defence

Uptime is defended at the racks and lost in the plant room. The BMS, the cooling sequences and the power chain are control systems with vendor tunnels and decade-old firmware, sitting metres from tenant workloads — an estate that belongs to facilities, and a defence that has to be engineered the way a plant's is.

Serving Saudi Arabia from our Riyadh office and our engineering hubs in Lahore and Karachi

Close view of circuit board components under directional light

Cybersecurity credentials

  • National Cybersecurity Authority, Saudi Arabia

    NCA registration

    Registered cybersecurity service provider

  • ISO

    ISO/IEC 27001

    ISO/IEC 27001 certified

Certifications & registrations

The blind spot

Facility control, the estate the IT floor never audits

We do not have an office in Saudi Arabia

We serve organisations there from our Riyadh office and our engineering hubs in Lahore and Karachi, travelling for the parts of an engagement that need to happen in person.

A data centre's security programme is usually excellent about the halls — hardened tenant boundaries, audited access, monitored everything — and silent about the systems the halls cannot exist without. The building management system, the chiller sequences and the switchgear controllers are procured by facilities, installed by contractors, serviced by vendors, and rarely appear in any security review, because they belong to a different department and speak protocols the IT tools do not read. Yet their consequence profile is the site's: the workloads survive a compromised tenant portal, and they do not survive a machine hall that loses cooling.

The structural fix is a boundary as formal as the ones between tenants. Facility OT gets its own zone with its own conduits: the BMS speaks to its controllers inside it, the monitoring the operations centre needs is published outward one way, and neither the corporate network nor any tenant network holds a route in. That line has an organisational half too — the facility estate needs a security owner, jointly agreed between the CISO and the facilities director, because an estate that belongs to everyone in principle is audited by no one in practice.

The estate

What actually runs the building beneath the racks

Four control estates, each with its own vendors, protocols and failure modes — and each a distinct entry in the attack surface.

01

The building management system

The supervisory layer over everything below — long-lived, vendor-managed, and historically installed with remote access as a convenience. Whoever holds the BMS holds a console over the building's physical behaviour, which is why its access list is the first thing worth reading.

02

Cooling control

Chiller plants, CRAC units and their sequencing logic, holding the variable a data hall tolerates least. The margin between normal and emergency is measured in minutes of thermal headroom, so this estate's protection is judged by what can reach it, not by what is installed on it.

03

The power chain

UPS controllers, generator sets and switchgear — equipment bought for electrical reliability whose management interfaces were an afterthought. The redundancy maths of N+1 assumes independent failures; a network that reaches every unit's management port quietly breaks that assumption.

04

Access and fire systems

Badge controllers, CCTV and suppression panels — the systems that decide who enters and what happens when something burns. They are networked, often onto the same facility segments, and a suppression discharge triggered or suppressed at the wrong moment is a physical event with the building's consequence scale.

Vendor access

A BMS vendor's session, from request to recorded closure

Facility vendors genuinely need access — the maintenance contract depends on it. The governed path gives them everything the job needs and nothing that outlives it.

A vendor session travels four stages — a request naming the task, one brokered gateway, recorded work, closure on a date — with a break-glass branch for emergencies that is separately credentialed and reviewed after every use.

  1. 01

    A request that names the task

    Access is requested against a work order — which systems, what work, which window — so the session has a defined shape before it opens.

  2. 02

    One brokered gateway

    The session runs through a gateway the operator controls, authenticated to a named technician — no standing tunnels from vendor equipment, and no route that skips the front door.

  3. 03

    Work watched and recorded

    The session is recorded while it lasts and scoped to the named systems, so what was changed during a maintenance visit is a matter of record rather than recollection.

  4. 04

    Closure on a date

    Access ends when the work order does; renewal is a decision, and the register of open vendor routes is reviewed on a schedule so nothing quietly becomes permanent.

Always available

The break-glass path

A chiller fault at three in the morning cannot wait for a work order, so an emergency path exists — pre-agreed, differently credentialed, and loudly logged, with a review of every use the next working day. The emergency is planned for; the audit trail survives it.

Monitoring

Listening to the plant room without leaning on it

The instruments that watch the IT floor — agents, scanners, automated response — are the wrong ones here. The facility estate is watched the way a plant is: from the wire, passively.

STAGE 01

Tap, never touch

Collection comes from mirror ports and network taps. No agent is installed on a controller, no scan is aimed at one — the monitoring must be incapable of causing the event it exists to catch. Visibility with a provably zero footprint on the estate.

STAGE 02

Learn the building's rhythm

Facility traffic is small and repetitive — the same controllers, the same polls, the same setpoint writes on the same cycle — which makes its baseline unusually honest and deviations unusually visible. A baseline against which a new conversation stands out.

STAGE 03

Alert on the new conversation

The event worth waking someone for is specific: a workstation reaching switchgear it has never addressed, a setpoint write from outside the BMS, a vendor session outside any work order. These flow into the same SOC view as the IT estate — one picture, different instruments behind it. Facility events reaching the SOC with the context to act.

STAGE 04

Respond with facilities, by hand

Response actions are rehearsed with the facilities team and executed by people, never automated — an automated containment that isolates the wrong segment is itself a cooling incident. What happens for each alert class is agreed before the first alert fires. A playbook the facilities team has walked, not just signed.

Our role

Our part in the plant room and the authority facilities keeps

  1. 01

    The boundary and the gateway

    We design and implement the segmentation between facility OT, corporate IT and tenant networks, and stand up the brokered vendor gateway — the two structures everything else depends on.

  2. 02

    Eyes on the estate

    We build the passive monitoring, tune its baseline with the people who know the building's rhythms, and wire its alerts into the SOC view your analysts already watch.

  3. 03

    Never between the operator and the building

    Nothing we deploy sits in the control path of cooling or power, and any change to a sequence or a setpoint is proposed to and executed through facilities engineering and the vendors' own service processes.

How the building runs — sequences, setpoints, maintenance — stays with your facilities engineers and their vendors; our work is to make sure the only hands that can reach those systems are the ones that are supposed to.

Common questions

What buyers ask

Is the building management system a realistic route into a data centre?+

Realistic enough that it deserves the assumption. Three properties make the facility estate attractive from an attacker's seat: it is reachable, because BMS platforms were historically installed with remote access for the maintenance contract and those routes outlive their reasons; it is soft, because the controllers and their operating systems are patched on facility timescales, not IT ones; and it is consequential, because the systems it commands decide whether the halls stay cold and powered. Whether an intruder's goal is the building itself or a foothold near more defended networks, the plant room offers a quieter path than the front door — which is the argument for giving it the boundary, the vendor governance and the monitoring the rest of the site already has.

How separate should facility OT be from the tenant IT it keeps alive?+

As separate as two tenants are from each other — that is the standard the boundary should borrow. Concretely: facility OT lives in its own zone, no tenant network and no corporate workstation holds a route into it, and the legitimate crossings are few and one-directional — monitoring data published outward to the operations centre, alarms outward to the SOC, and nothing initiated from outside reaching inward to a controller. The subtle crossings deserve the most attention, because they are the ones drawings omit: the DCIM platform that reads facility data and lives on the corporate network, the shared jump host from a commissioning project, the vendor laptop that visits both sides. For a colocation operator the boundary is also a commercial fact: tenants are entitled to assume that no path leads from another tenant, through the building, to them.

Can cooling and power controllers be monitored without destabilising them?+

Yes, provided the monitoring never transmits toward them. The safe architecture collects from mirror ports and taps, so the controllers never see the monitoring at all — it cannot destabilise what it cannot touch. What must be kept off this estate are the IT instincts: an active scanner probing a chiller controller, an agent pushed to a BMS server pinned on an old operating system by vendor qualification, an automated response that isolates a segment and takes the sensors with it. Passive collection costs something — it observes rather than interrogates, so device internals are inferred from traffic and records instead of measured directly — but on equipment where a probe can hang the device, that trade is not close. The facility estate rewards it doubly, because its traffic is so repetitive that passive baselining catches the anomalies that matter.

Review your facility estate with us

A facility review that lists every control system beneath your halls, who can reach each one today, and the first boundary to build between facility OT and everything else.

What happens next

  1. 1We reply within one business day.
  2. 2A 30-minute scoping call with a senior consultant.
  3. 3A written proposal with scope, team, and timeline.

Or reach us directly

By submitting this form you agree to our Terms of Use and Privacy Policy.