MQTT vs OPC UA: The Industrial Data Showdown
Compare the Two Protocols Defining Modern Industrial Data - and Learn When to Use Each
Almost every industrial data architecture eventually hits the same fork: MQTT or OPC UA? In this session, Kristopher Sandoval and Kudzai Manditereza from HiveMQ put both protocols head to head — how each moves data, where each breaks down, and what the architecture actually looks like when you stop choosing between them.
Kudzai made the case for MQTT. Kristopher made the case for OPC UA — and backed it up with a live demo using FlowFuse's OPC UA Certified Node in a real deployment scenario: a multi-site gene therapy manufacturing flow spanning San Francisco, Funabashi, and Hanover, handling ~4,500 topics per second across sites.
The session covered:
- What MQTT and OPC UA each do well, and where each falls short
- Why the "winner" depends on your use case, not the marketing
- How MQTT's pub/sub architecture enables data reuse and reduces attack surface in IT/OT environments
- When OPC UA's bidirectional, determinative control is a hard requirement — and when it isn't
- How teams combine both protocols instead of choosing one
- Practical patterns for building a Unified Namespace that doesn't lock you in
The audience weighed in live: most attendees were already running both protocols in production. The biggest open question wasn't which protocol to pick, but how to use both together — and when each one earns its complexity.
Watch the recording to see the full debate, the live demo, and the Q&A — including questions on contextualization across assets with OPC UA, handling real-time, alarm, and historical data streams in parallel, and whether OPC UA is a technical necessity or a regulatory artifact in pharma and life sciences.
Still figuring out where MQTT ends and OPC UA begins?
If this session clarified the architectural question but left you wondering how to actually bridge both protocols in your environment, FlowFuse is built for exactly that — with a certified OPC UA node, native MQTT support, and the ability to run logic at the edge where the data lives.
