The outage now reaches the control room before the phone rings

The client is a power utility operating transmission and distribution assets across its service territory. Like most utilities, its monitoring estate had grown system by system. SCADA reported into its own tool. The metering head-end reported into another. Field sensors fed yet another. Each system did its job. None of them talked to each other, and no shared picture of the grid existed anywhere.

The cost of that fragmentation showed up in the control room. Operators often learned about faults from inbound customer calls rather than from the network itself. Maintenance crews responded to failures instead of getting ahead of them. And every reporting cycle, staff pulled exports out of each system and assembled the regulator's reliability reports by hand, with no clean trail behind the figures submitted.

The utility engaged Kentro to build a central telemetry platform. The brief was direct: ingest every sensor feed into a single stream, detect outages as they happen, put analytics on top for predictive maintenance, and generate compliance reporting straight from the data. The utility's own team would need to operate and extend the platform, so it had to be built on open, proven infrastructure.

The utility's monitoring estate was not broken. It was fragmented, and the fragmentation was the problem. Every system worked in isolation, so the grid as a whole was effectively running blind.

• Siloed telemetry: SCADA, the metering head-end and field sensors each fed a separate system with its own tooling and its own data model. There was no shared view of the grid and no common record for any asset on it.
• Outages announced by customers: With no correlated view of the network, the control room often learned about faults from inbound calls. Customers were, in practice, part of the monitoring system.
• Reporting assembled by hand: Reliability reports for the regulator were compiled manually from exports pulled out of each system. The process was slow, error-prone and difficult to audit after submission.

Kentro built a layered platform that turns raw grid telemetry into operational decisions. A streaming backbone sits at the bottom, an analytics layer in the middle, and reporting on top. The whole platform runs on Kubernetes on the utility's private infrastructure, on open components the utility's own engineers can operate and extend.

• A streaming ingestion backbone: Kafka pipelines pull SCADA feeds over OPC UA and Modbus, meter events from the metering head-end, and field sensor readings over MQTT into a single stream. Every source that once fed its own silo now feeds the platform.
• A time-series store on a common asset model: TimescaleDB holds full sensor histories keyed to a shared grid asset model, so every transformer, feeder and switch has a single record that operations, maintenance and compliance all reference.
• Real-time outage detection: Flink jobs correlate meter last-gasp events with feeder telemetry to locate faults and open incidents automatically. Live network state surfaces on Grafana views in the control room.
• Predictive maintenance models: Python models trained on load and temperature histories flag transformers and switchgear for inspection before they fail, and feed maintenance planners a ranked inspection queue.
• Automated compliance reporting: Scheduled jobs assemble regulator-ready reliability reports straight from platform data, with a full audit trail behind every figure submitted.

The client asked us not to publish figures, and the honest measure of this platform is operational anyway. What changed is how the utility's teams work day to day.

• The control room sees faults first: Outages now appear on operator screens as they happen. Customer calls confirm incidents instead of announcing them.
• Maintenance runs on evidence: Crews are dispatched against model-flagged equipment instead of waiting for failures in the field. Inspection effort goes where the data points.
• Reporting without the manual run: Reliability submissions are generated from platform data, replacing the spreadsheet exports that staff used to compile by hand. Every submitted figure can be traced back to source telemetry.
• A single asset record across teams: Operations, maintenance and compliance now work from the same record for every asset on the grid, on a platform the utility's own team operates.