The backbone of an industrial telemetry platform: it receives, decodes and stores the transmissions of trackers and on-board computers installed in heavy machinery, and exposes everything through APIs to the embedded applications and to the client portal. The database holds over 700 million telemetry records.
The challenge
Heavy machinery works far from any office — on construction sites, in mines, in industrial plants, out in the fields. The data that matters to the manager (engine hours, fuel use, position, speed, faults, actual equipment usage) is born locked inside each manufacturer's hardware, on proprietary buses, and historically only reached the manager after the problem had already cost availability. And it is not data that arrives at a civilised pace: hundreds of machines transmit continuously, 24 hours a day, over an unstable mobile connection, in different binary protocols depending on the tracker installed.
The solution
Catus is the core of that platform. A dedicated TCP gateway answers the trackers' connections directly on a socket, decodes the manufacturer's protocol and hands the transmission to a processing queue — which lets it absorb spikes without losing data and without dragging down the rest of the system. On top of that raw stream sits the domain model, with more than 220 entities, which turns transmissions into useful information: equipment status, sensor readings, CAN bus data, G-force events, geofence entries and exits, refuelling, shifts and productivity reporting. The internal back office handles what the customer never sees: registering and configuring equipment, devices, SIM cards, sensors, peripherals and cameras, plus remote diagnostics for every terminal in the field. And a 25-endpoint API serves the embedded application on the machines exclusively, including offline synchronisation and remote application updates delivered in chunks — necessary because a connection in the field rarely survives a whole file. Completing the platform is a scheduled-jobs service, with 22 periodic routines, that fires alerts, consolidates reports, consumes a Kafka topic for integration with an external partner and even monitors the cost of its own AWS infrastructure.
Results
More than 700 million telemetry records accumulated, with roughly 2,200 machines and 1,100 devices tracked for over 250 companies. The whole platform runs in containers on Amazon ECS, with four independent services published through an automated pipeline.