Solace in Manufacturing: The Enterprise Backbone Above Plant MQTT Brokers

Solace in manufacturing is best understood as the enterprise event backbone that sits above a plant’s MQTT brokers, not as the broker next to the machines. Its event broker can accept MQTT clients directly, but what Solace adds that a plant broker does not is routing events between many sites, regions and clouds, guaranteed delivery into ERP, MES and quality systems, protocol translation for IT applications (JMS, AMQP, REST) and design-time governance. Solace’s own manufacturing page names Unified Namespace, supply chain, Industry 4.0, servitization and workforce augmentation as its use cases, and lists ten manufacturing customers.

This page places Solace on the ISA-95 stack, summarises who uses it and for what, and sets out the limits a plant team should test before committing. It is part of the Solace cluster, an independent overview. Joltek has no commercial relationship with Solace.

Where Solace fits in manufacturing: above the plant broker

A plant needs two different kinds of messaging, and they rarely come from one product. Close to the machines, the job is to collect state from PLCs, drives and instruments, publish by exception, and give SCADA, MES and dashboards one shared view of “what is true now”. That is the plant unified namespace, usually on an MQTT broker, often with Sparkplug B. Above the plant, the job changes: move selected events between sites and into enterprise systems, make sure a production confirmation reaches SAP exactly when the WAN comes back, and stop one slow consumer from holding up everyone else.

Solace is built for the second job. Its April 2026 UNS blog makes the same distinction in Solace’s own words: “Don’t abandon UNS, evolve it” and “Don’t force every use case into MQTT”. It argues that an event mesh supplies “durable fan-out”, “consumer isolation” and “secure, policy-driven routing” on top of the shared view a UNS provides.

Why not one broker for everything? Because the two layers fail differently. A plant broker that goes down stops the site’s dashboards and MES feeds, so it belongs inside the plant, run by people who can reach it at 3 a.m. An enterprise backbone that loses a WAN link must hold events and deliver them later without the plant noticing. Splitting the jobs lets each layer be sized, secured and supported for its own failure mode. The full multi-site picture, including the DMZ, is on multi-site edge to cloud, and the routing mechanism is explained on the event mesh page.

What Solace says it does for manufacturers

Solace’s manufacturing page claims the platform distributes real-time information between IT and OT assets, moves operational data between machines, lines and sites, works across cloud, on-premises and edge, and can “elastically scale your system to manage millions of devices and billions of real-time events”. Those are Solace’s statements. No public benchmark ties them to a manufacturing deployment.

In practice the event flows a manufacturer puts on an enterprise backbone look like this (illustrative):

1. Master data down. A new material or recipe version is published once from ERP or PLM and delivered to every site’s MES that subscribes to it.
2. Production orders down, confirmations up. Orders flow from ERP to the line; goods receipts, scrap and confirmations flow back. Each confirmation is a guaranteed message on a queue, so a WAN outage delays it rather than losing it.
3. Quality events sideways. A failed in-process check publishes a hold event that MES, the quality system and the warehouse all receive, so the affected pallets stop moving before they ship.
4. Machine state up, summarised. Not every tag. A site publishes line state, counts and alarms at the granularity the enterprise needs, and keeps the high-frequency data in its own UNS and historian.

The fourth point is where many designs go wrong. Pushing every tag change from every line to the enterprise layer costs WAN bandwidth, broker capacity and money, and almost nobody consumes it. Decide which topics cross each boundary, and treat that list as a design document.

Named Solace manufacturing customers

Solace lists Absolics, Airbus, Bosch, Danone, Groupe Rocher, Heineken, JDE Peet’s, Renault Group, Sonos and Stellantis on its manufacturing page, as of October 2026. The table summarises what is actually public for each. Everything in it is Solace’s or the customer’s statement, not ours.

Customer	What is public	Source
Danone	An event mesh across more than 100 production sites and 300 distribution centres, integrating MES, WMS, SAP and IoT. Digital manufacturing manager Mathias Ollion described IT-to-OT data as previously “locked into applications or point-to-point interfaces that introduced latency”. The manufacturing page calls the programme “Cortex”	Danone story
Bosch	Production master data distribution, production order and confirmation workflows, and an OT-IT data provisioning layer. Results are stated qualitatively	Bosch story
Heineken	An event-driven “Digital Backbone” integrating thousands of applications across operations in more than 190 countries. No brewery-level OT detail is published	Heineken story
JDE Peet’s	Real-time distribution through SAP Integration Suite, advanced event mesh	Solace on SAP
Absolics	An SKC subsidiary building a U.S. glass-substrate factory; a cloud-based event mesh joining manufacturing, management and product development applications (September 2024)	Absolics release
Apex Lubrication	Solace Agent Mesh inside the OmniView conveyor-monitoring platform with Mighty Lube, using on-premises LLMs and routing to ERP and CMMS (June 2026)	Apex Lubrication release
Airbus, Renault Group, Stellantis, Groupe Rocher, Sonos	Listed by Solace; short descriptions only, no public architecture	manufacturing page

Read the pattern, not the logos. The detailed stories are about enterprise integration: master data, orders, confirmations, distribution. None of them describes Solace replacing a plant’s SCADA or its tag-level MQTT broker. That is consistent with the backbone positioning, and it is the honest reading of the evidence. Apex Lubrication is the first public factory-floor AI reference; the Agent Mesh page covers that product.

Plant broker or Solace backbone: how to decide

Situation	Plant-class MQTT broker	Solace at the enterprise layer
One site, SCADA, MES and dashboards share state	Usually enough on its own	Over-specified
Sparkplug state handling is central (Ignition-heavy)	Eclipse-listed brokers such as Chariot or HiveMQ	Possible, but test retained and Will behaviour
Five or more sites, one data strategy	One broker per site, still needed	Links the sites and the enterprise
WAN links drop and events must not be lost	Local buffering at the edge	Guaranteed queues and replay across sites
ERP, QMS or data platforms need JMS, AMQP or REST	Bridges or custom code	Native on the same broker
SAP is the integration hub	Point connectors	Often arrives as SAP advanced event mesh
Long retention and analytics replay	Not its job	Usually paired with Kafka; see Kafka integration

The decision is rarely “Solace or HiveMQ”. It is “what runs in each plant, what connects the plants, and who operates each layer”. For the head-to-head with the most common plant broker, see Solace vs HiveMQ. For the vendor-neutral version of the broker choice, see choosing a UNS broker. Our IT/OT integration work starts from exactly this layering question.

Limits and open questions

• MQTT deviations. The MQTT documentation states that published QoS 2 messages are converted to QoS 1, that only QoS 0 shared subscriptions are supported (higher requests are downgraded), and that retained messages are a Controlled Availability feature. Detail is on Solace MQTT.
• Will messages on failover. The same page says the broker does not send Will messages when a session ends because of a broker restart, an HA failover or a Message VPN shutdown. Sparkplug death certificates are Will messages, so this matters.
• Sparkplug status. A Solace community post from August 2023 reported broker 10.5 passing the Eclipse Sparkplug test kit. Solace is not on the Eclipse compatible list as of October 2026. See Solace Sparkplug.
• The free edition is not for production. The Standard Edition is documented as “Production Deployable: No”, with 1,000 connections and a 10,000 messages per second ingress cap. A plant pilot that will go live needs Enterprise or Solace Cloud. See Standard vs Enterprise.
• No public pricing for Enterprise, Solace Cloud or Agent Mesh, so a multi-site business case needs a quotation early.
• Thin public manufacturing detail. Ten logos, three detailed stories, and none at tag level. Ask for a reference that matches your architecture.
• Skills. Message VPNs, queues, dynamic routing and Event Portal assume an IT integration team. Decide who on the OT side will own the site end.

How to evaluate Solace for manufacturing

• Which layer are we buying for: plant broker, enterprise backbone, or both? Write it down before the demo.
• Which topics cross from each site to the enterprise, at what rate, and who consumes them?
• What happens to a production confirmation when the WAN is down for four hours? Show us, on our topology.
• How do our Sparkplug edge nodes and host applications behave on a Solace broker, including HA failover and Will messages?
• Do we need QoS 2 anywhere, or shared subscriptions above QoS 0?
• Will Solace arrive directly or as SAP advanced event mesh, and who supports which layer?
• Where do the site brokers sit relative to the DMZ, and which side opens each connection?
• Which reference customer runs a comparable number of sites with a comparable OT stack?
• What does the three-year cost look like for our site count, and what is the cost of running a plant broker beside it?

If the topology itself is not yet documented, start with an OT network assessment; the broker design follows from the zones and conduits, not the other way round.

Deeper

• Solace Unified Namespace: a single-site UNS versus a multi-site event mesh, and how to design the namespace that crosses them
• Solace Sparkplug: Sparkplug B 3.0, what compliance means, the 2023 test versus the Eclipse list, and state across a mesh
• Solace and Ignition, HighByte: integration patterns for the two listed OT partners
• Solace Edge to Cloud: many plants, intermittent WAN, store-and-forward and DMZ placement

Back to the Solace hub, or across to Protocols and APIs for the full protocol list and to SAP advanced event mesh for the SAP route. For MQTT fundamentals, our guide to MQTT in manufacturing covers the protocol itself.