Techbanq

Industrial security · segmentation

Zones for a network that grew flat under a running plant

Plant networks go flat one convenient connection at a time, and no plant can be re-cabled while it runs. The way back is sequenced: the highest-consequence unit gets its zone first, every boundary rule proves itself in monitor mode against real traffic, and enforcement happens only inside windows operations has agreed.

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

Market
Saudi Arabia
Written for
Plant Manager, CISO
Decision
Introduce zones and conduits into a brownfield plant network without stopping production

Credentials for energy and industrial delivery

  • Saudi Aramco

    Aramco CCC

    CCC certified

  • ISO

    ISO/IEC 27001

    ISO/IEC 27001 certified

Certifications & registrations

The sequencing

The order in which a flat estate earns its zones

Segmenting everything at once is a project that never finishes; segmenting by consequence puts protection where the loss would be largest, early.

STAGE 01

Baseline before drawing

Weeks of passive capture establish what actually talks to what. Every zone proposal starts from this record, because a boundary designed from the official drawing blocks the flows the drawing forgot. A flow map that outranks the archived diagrams.

STAGE 02

Rank the units by what their loss costs

Operations names the process units whose compromise or outage hurts most — and the safety systems, which get their own zone on principle. That ranking, not network topology, decides which boundary is built first. A build order the plant leadership has endorsed.

STAGE 03

Prove each boundary in monitor mode

The boundary runs first as an observer, counting what it would have blocked. Each would-be block is reviewed with operations and named legitimate or not, so the rule base is corrected on paper instead of in production. A rule set that demonstrably blocks nothing the process needs.

STAGE 04

Enforce inside an agreed window

Enforcement flips on during a maintenance window with operators watching the process indicators and a rehearsed way back. Then the next unit in the ranking gets the same treatment. Protection landing unit by unit, with production never interrupted.

The vocabulary

Purdue is a map you talk over, not a law you enforce

The Purdue model earns its place in every segmentation conversation for one reason: it gives the plant and the security team a shared vocabulary. Level 1 devices, a level 3 historian, a boundary between 3 and 4 — everyone in the room means the same thing, and that alone prevents half the misunderstandings these projects generate. What the model is not is a compliance target. Real estates violate its clean layers everywhere: historians straddle levels, vendor systems arrive with their own flat networks, and a decade of commissioning shortcuts has cross-wired what the diagram separates.

The brownfield discipline is therefore to use the model for where boundaries should sit, and the plant's reality for where they can. You cannot re-cable a running plant, so zones follow the switches, cabinets and windows that exist: sometimes a boundary lands on a VLAN and a firewall rather than on new cable, sometimes two units share a zone for a year because separating them needs a turnaround. An imperfect zone enforced this quarter protects more than a textbook zone scheduled for a shutdown that keeps slipping — and the design records the debt explicitly, so the next turnaround closes it rather than forgetting it.

One rule's journey

A firewall rule's route from proposal to a boundary production trusts

The same path applies to the first rule and the five-hundredth. Its slowness is deliberate: on a plant network, an untested block is an outage with a start time.

A firewall rule travels four stages — proposed from observed traffic, run in monitor mode, reviewed with operations, enforced in a maintenance window — with a rollback branch that reverts it inside the same window.

  1. 01

    Proposed from observation

    The rule is written from captured traffic — named sources, named destinations, the protocol actually in use — never from an assumption about what should be talking.

  2. 02

    Run silent, counting

    In monitor mode the rule logs what it would have blocked, for long enough to cover the slow rhythms — the weekly report job, the monthly vendor sync.

  3. 03

    Reviewed with operations

    Every would-be block is walked through with the people who run the process, and each is named legitimate traffic to permit or a finding to investigate.

  4. 04

    Enforced in a window

    The switch to blocking happens in a maintenance window, with process indicators watched and the change recorded with its reason and owner.

Always available

Revert inside the window

Every enforcement carries a rehearsed rollback that fits inside the same window. If an indicator moves, the rule comes off before the window closes — the plant never exits a window worse than it entered.

Our role

Boundary discipline the operations team can keep after we leave

  1. 01

    Rules written in plant language

    Each rule names stations, devices and protocols the way the plant names them, so a control engineer can read the rule base and recognise the estate — a rule base only a firewall specialist can read decays the day the specialist leaves.

  2. 02

    Exceptions with expiry dates

    Commissioning jobs and vendor visits get temporary rules that carry a reason and an end date, and a reconciliation routine sweeps the rule base against the design on a schedule — the mechanism that stops the network drifting flat again.

  3. 03

    The routine handed over deliberately

    Rule review, exception approval and the reconciliation sweep are documented and run jointly until your team runs them alone, because segmentation is only permanent if its upkeep belongs to the people who stay.

Which flows count as production-critical, and when a window opens, are operations' calls — approved through the plant's engineering authority where the control path is touched; we design, prove and enforce within them.

Common questions

What buyers ask

Do we need the full Purdue model before we can call the plant segmented?+

No — and chasing it can delay real protection. Purdue is most useful as the shared language in which the plant and the security team discuss where boundaries belong; as a build specification it assumes a greenfield nobody has. The honest question is not whether your network matches the diagram but whether the highest-consequence units sit behind enforced, monitored boundaries, whether the safety systems have their own zone, and whether the paths between business and control networks are few, named and watched. An estate can pass that test while violating the textbook layering in a dozen places, and it is far better protected than one that postponed everything for a perfect redesign.

Can a brownfield plant reach real segmentation without re-cabling?+

Substantially, yes. Most of the protection comes from boundaries that existing infrastructure can already carry: VLANs on the switches that are there, firewalls at the few points where traffic naturally concentrates, and access-control lists that turn a shared switch into a policed one. What cable-free segmentation cannot always deliver is separation inside a single unmanaged segment — two devices on the same dumb switch cannot be firewalled from each other — so those cases go on a list, priced, and scheduled for the next turnaround when hardware can move. The design distinguishes honestly between boundaries enforced now, boundaries pending hardware, and boundaries not worth their cost, which is a statement most flat networks have never had.

What stops a new boundary from blocking traffic the process depends on?+

Three gates, in order. First, rules are written from captured traffic rather than from expectations, so the flows that exist are known before anything is drawn. Second, every boundary runs in monitor mode long enough to cover the slow rhythms — weekly batch jobs, monthly vendor synchronisations are the classic surprises — and each would-be block is reviewed with operations by name. Third, enforcement happens only inside a maintenance window with process indicators watched and a rollback rehearsed to fit inside the same window. The residual risk is never zero, which is exactly why the third gate exists: if something moves, the rule comes off before the window closes, and the finding improves the rule instead of stopping the plant.

Next step

Plan the first zone with us

A working session over your current network drawings and a traffic capture, ending with the first zone chosen by consequence and its enforcement window pencilled in.