Skip to main content

General

Core Concepts

Understanding the foundational principles of Proxus: Single Source of Truth, Stateless Edge, and Guaranteed Delivery.

Proxus is built on a few non-negotiable principles that differentiate it from traditional SCADA or IoT platforms.

1. Single Source of Truth (SSoT) Core Philosophy

In many industrial environments, data models are duplicated across PLCs, SCADA, Historians, and MES systems. A change in one requires manual updates in all others.

Proxus changes this. You define your data model (Assets, Tags, Units, Scaling) once in the Central Management Core. This definition is then automatically propagated to:

  • The relevant Edge Gateways (for data collection).
  • The Unified Namespace (for MQTT topics).
  • The Time-Series Database (for storage).
  • The Dashboard UI (for visualization).

2. Stateless Edge Runtime

The Edge software is designed to be disposable. It holds no persistent configuration state on the disk.

Boot

The Gateway starts up and connects to the Central Core via NATS.

Handshake

It identifies itself using its unique Gateway ID and requests its configuration.

Configuration

The Central Core pushes the latest driver settings, tag lists, and logic scripts.

Execution

The Gateway begins polling devices and running logic immediately.

Benefit: If a gateway hardware fails, simply replace the box and install the generic Docker image. It will auto-configure itself in seconds.

3. Guaranteed Delivery (Jetstream)

Industrial networks are unreliable. Proxus uses NATS Jetstream to ensure data integrity.

  • Buffering: If the connection to Central drops, the Edge buffers data locally on disk.
  • Replay: When connection is restored, data is replayed in exact order.
  • Backpressure: The system handles bursty data loads without crashing.

4. The Unified Namespace (UNS)

Proxus acts as a central nervous system, organizing all your data into a semantic hierarchy based on ISA-95.

Enterprise / Site / Area / Line / Cell / Machine / Tag

Consumers (ERP, MES, AI) subscribe to the nodes they care about, rather than creating point-to-point connections to every PLC.