OT/IT convergence: lessons from African mining
Mining operations across the region are connecting plant systems to enterprise networks faster than they are securing them. The engineering lessons are consistent, and they are not primarily about technology.
- Convergence programmes fail on operating-model boundaries, not on network design.
- Availability, not confidentiality, is the primary security objective on the plant floor.
- Passive discovery beats asset registers that were accurate three shifts ago.
- Value shows up first in unplanned-downtime reduction, not in dashboards.
The boundary problem
On most sites, control systems belong to engineering and networks belong to IT. Convergence puts a shared asset between two functions with different incentives: engineering is measured on production availability, IT on security and change discipline. Unless the operating model is renegotiated explicitly — who may patch, during which window, with whose approval — the technical architecture will be quietly bypassed within a quarter.
The sites that get this right create a joint operations forum with a single accountable owner for the converged zone, and they write change authority down. It is unglamorous work that determines whether the rest holds.
Security objectives invert on the plant floor
Enterprise security prioritises confidentiality; industrial environments prioritise availability and safety. A control that halts a mill to contain a suspected intrusion may cause more harm than the intrusion. Segmentation, monitoring and manual containment procedures therefore outperform automated blocking in the first phases of maturity.
Purdue-model segmentation remains the pragmatic reference. Most sites we assess have flat networks below Level 3 and a firewall that was configured during commissioning and never revisited.
- Passive asset discovery before any active scanning.
- Unidirectional or brokered data flows out of the control zone.
- Separate remote-access path for vendors, time-bound and recorded.
- Incident playbooks written jointly with plant engineering.
Where the value actually appears
The business case is usually written around predictive maintenance and reporting. In practice the first measurable return comes from something simpler: reliable, timely visibility of unplanned downtime causes across shifts. Once that data is trustworthy, condition monitoring and maintenance optimisation follow with far less resistance because the operations team already believes the numbers.