CTS Logger

CTS Logger is an industrial data-acquisition application that has run in production for six months without losing a single reading. It speaks two transports at once — RS485/COM serial and WiFi/HTTP — to more than 70 concurrent devices on a factory floor.

The problem

The plant had a mixed fleet of instruments: legacy meters that only speak Modbus-style serial over RS485, and newer IoT devices that report over HTTP. Existing tooling assumed one transport or the other, and polling 70+ devices naively meant duplicate commands piling up whenever a device responded slowly — which on a shared serial bus means collisions, garbage frames, and lost data.

The design

The application is an Electron shell around a Node.js acquisition core, so the operators get a desktop UI while the engine runs the same on any machine.

Dual transport, one pipeline. Serial (RS485/COM) and network (WiFi/HTTP) devices are abstracted behind a common device interface. The acquisition scheduler doesn't know or care which wire a reading arrives on; transport adapters handle framing, retries, and timeouts per medium.

Command queue with auto-deduplication. Every outbound request goes through a central queue. If a poll for a device is already pending, an identical request is deduplicated instead of enqueued — so a slow device can never cause a backlog of redundant commands, and the serial bus stays clean under load.

Designed for not losing data. Readings are persisted before they are acknowledged, writes are append-first, and the pipeline recovers from device disconnects, bus errors, and process restarts without dropping buffered readings.

Results

• 70+ devices polled concurrently across two transports
• 6 months of production uptime
• Zero data loss over that entire period
• Operators add and reconfigure devices without touching code