United Manufacturing Hub

United Manufacturing Hub (UMH) is an open-source industrial data management platform developed by UMH Systems GmbH in Aachen, Germany. It solves a specific problem: manufacturing plants run dozens of incompatible protocols and siloed systems, making it nearly impossible to get a single, real-time view of production data. UMH packages proven open-source components into one deployable stack that creates a Unified Namespace (UNS) — a centralized, ISA-95-structured data layer where every machine, sensor, and enterprise system publishes and subscribes to the same event stream. The company raised EUR 5M in January 2026, is ISO 27001 certified, and counts Hipp, Roche, Boellhoff, and E.ON among its customers.

Architecture

UMH is built around three layers. The Data Infrastructure handles connectivity, brokerage, and storage. Protocol converters (supporting OPC-UA, Siemens S7, Modbus TCP, Ethernet/IP, Allen-Bradley, REST, SQL, and MQTT among 70+ input types) pull data from the shop floor. That data flows into the UNS via HiveMQ (MQTT broker) and Redpanda (Kafka-compatible streaming), where Benthos stream processors handle transformation, routing, and enrichment. TimescaleDB stores historical data, and Grafana provides dashboards and alerting.

The Device and Container Infrastructure automates deployment on Kubernetes. UMH ships as a Helm chart that orchestrates all components, with automatic recovery and rolling upgrades. A lighter option ("UMH Lite") runs as a single Docker container for smaller sites or evaluation.

The Management Console is a web application for configuring connections, browsing the UNS, tracing data lineage from PLC to database, and monitoring instance health across sites — including instances behind DMZs.

The older "UMH Classic" version used Apache Kafka and required ZooKeeper. The current architecture (UMH Core) replaced Kafka with Redpanda, reducing operational overhead while maintaining Kafka API compatibility.

Key features

• Unified Namespace with ISA-95 data model, linking every layer of the automation pyramid into a single event-driven source of truth
• 70+ protocol inputs and 80+ outputs including OPC-UA, S7, Modbus TCP, Ethernet/IP, MQTT, REST, SQL, and cloud connectors for AWS, Azure, and GCP
• Management Console for multi-site deployment, configuration, UNS browsing, data lineage tracing, and health monitoring
• Pre-built Grafana dashboards for OEE monitoring, energy tracking, and alarm management
• Benthos-based data flow engine for in-stream transformation, filtering, and routing without writing code
• One-click upgrades via the Management Console, no Kubernetes expertise needed for day-2 operations
• Node-RED integration for custom data flow orchestration and protocol bridging
• Open-source core under a permissive license with a free-forever Community Edition

Getting started

The Community Edition is free and includes all core components. Deployment takes roughly 15 minutes: the Management Console provisions a Kubernetes cluster (or a single Docker container for UMH Lite) and installs the full stack. For teams without Kubernetes experience, UMH manages the cluster lifecycle automatically. The Enterprise License adds multi-instance deployment with DMZ support, RBAC, enterprise SLA with four severity levels, LTS versions, and optional add-ons like 24/7 support and custom protocol converters. Enterprise pricing is not published — contact sales. UMH also maintains an international network of certified system integrators who handle on-site implementation.

Limitations

• Community Edition caps at 5,000 tags per site. Larger deployments require the Enterprise License, whose pricing is not public, making cost estimation difficult for budgeting.
• Kubernetes dependency remains a barrier. While the Management Console abstracts much of the complexity, troubleshooting production issues (pod crashes, persistent volume claims, network policies) still requires Kubernetes knowledge that many OT teams lack.
• Smaller ecosystem than general-purpose IIoT tools. With ~360 GitHub stars, UMH has fewer community-contributed plugins, tutorials, and Stack Overflow answers than Node-RED (22K+ stars) or ThingsBoard (21K+ stars).
• No built-in analytics or ML pipeline. UMH handles data collection, brokerage, and storage but does not include predictive maintenance, anomaly detection, or machine learning capabilities — these must be added via external tools.
• Legacy serial protocol support (Modbus RTU, PROFINET, HART) requires additional gateway hardware or third-party converters; the built-in protocol converters cover TCP/IP-based protocols only.