Niagara
Niagara runs millions of BMS systems worldwide. Integrar IoT connects to JACEs via Fox protocol.
Key Metrics
1M+
JACEs Worldwide
Fox Protocol
Native
Auto-Discovery
Multi-Vendor
Compatible
What the Niagara Framework Actually Is
Created by Tridium (now part of Honeywell) in the late 1990s, Niagara is not a single product but a Java-based software platform and licensing model that turns a commodity appliance into an open integration server for buildings and industrial sites. Its claim to fame is independence: one Niagara station can host drivers for dozens of vendor protocols - BACnet, LonWorks, Modbus, SNMP, OPC, and proprietary DDC dialects - and present them all through a unified data model and HTML5 user interface.
Because the platform runs on both hardened field controllers (JACEs) and supervisory servers (N4 Supervisors), the same software stack scales from a single rooftop unit in a strip mall to a campus-wide enterprise BMS. That consistency is why property owners standardize on Niagara: the skills, the graphics, and the tag naming carry from site to site regardless of which HVAC vendor won the last retrofit.
Niagara 4 Stations
A station is the running instance of the Niagara runtime - the engine that hosts drivers, control logic, alarms, histories, and the web server. Stations are configured in Workbench, then deployed to a JACE or supervisor where they run continuously.
JACE Controllers
JACE (Java Application Control Engine) is the embedded hardware family that runs stations at the edge. JACE-8000 and JACE 9000 units pack a small Linux host, a Niagara runtime license, and onboard I/O options into a DIN-rail enclosure that can run a whole building's integration without a server room.
Fox Protocol
Fox is Niagara's own encrypted remote-access and peer-to-peer protocol. Integrar IoT speaks Fox natively to read points, subscribe to alarms, and retrieve histories from any station without vendor engineering tools.
Drivers, Networks, and the Baja Component Model
Niagara's extensibility comes from two design decisions. First, every protocol connection is a software "driver" that plugs into the station, so adding a new device type means adding a driver module, not rewriting logic. Second, the platform is built on the Baja component model, a formal specification (also published by Tridium) that describes how software components expose points, properties, and services to each other.
| Layer | Component | What It Does |
|---|---|---|
| Device layer | Proxy points | Map a field register or BACnet object into a station point that can be alarmed, trended, or written |
| Network layer | Drivers / buses | Manage polling, transport, and device discovery for each protocol (BACnet, Modbus, LonWorks, SNMP, OPC, and more) |
| Station services | Alarm, history, scheduling | Central services that any point can attach to, giving uniform alarming and reporting across vendors |
| Presentation | Workbench / Web UI | HTML5 graphics and tag-based navigation rendered from the same station data |
How a Driver Talks to the Field
Each driver contains a device-discovery wizard that reads the actual network - listing BACnet devices that respond to Who-Is, LonWorks devices by Neuron ID, or Modbus nodes on a configured serial or TCP path. Discovered points are then bound as proxy points, and the driver's poll engine keeps them fresh at configurable intervals. Because the proxy point is a normal Baja component, a chilled-water valve bound from a Modbus slave behaves identically to one bound from a native BACnet actuator: it can be alarmed, scheduled, or exposed to the cloud through the same services.
The practical consequence for integrators is speed. A technician who knows the protocol map can bring an entire mechanical floor online in hours, and the point database is self-documenting because every proxy point carries its source address and engineering unit in its properties.
Why Multi-Vendor Integration Is the Whole Point
The most common reason buildings end up with a Niagara station is vendor lock-in on the way out, not in. A campus that has grown over twenty years will have Trane controls in one wing, Johnson Controls in another, Schneider in a third, and a dozen standalone controllers for special equipment. Niagara's value is that it gives those systems one shared namespace, one alarm console, and one graphics package - without ripping out any of them.
One Station, Many Protocols
A single JACE routinely hosts BACnet/IP, Modbus TCP, and a proprietary serial protocol simultaneously. Point names and units are normalized to the station, so an analytics layer only needs to understand the station, never the vendor.
Beyond HVAC
Niagara drivers extend to lighting, metering, elevators, generators, and power quality, letting facilities teams manage a true IoT estate instead of a set of siloed trades.
Best Practices When Extending a Niagara Site
- Keep the station licensed and patched: Niagara features are license-gated, and security patches are delivered through the software maintenance program. An unlicensed or unpatched station silently loses functionality and becomes an easy target.
- Respect the alarm and history hierarchy: attach histories at the point level with realistic intervals, and use the station's alarm service rather than burying alarms in graphics.
- Plan the station's IP security: restrict Fox and Web access to management VLANs, change default credentials, and terminate TLS rather than exposing the station to the office network.
- Document before you expand: a clear proxy-point naming convention (location-space-device-point) saves months of confusion when analytics or a new BMS vendor joins the project.
For organizations connecting Niagara to higher-level IoT systems, the important design principle is to treat the station as the source of truth for field data and to add a cloud or analytics layer only as a consumer - reading through Fox or a supervisory REST interface rather than reaching into the field network from outside.
Connect Your Niagara Fleet to IoT Analytics
Integrar IoT reads points, alarms, and histories from any JACE or supervisor and pipelines them into your analytics and cloud platforms - without touching the live controls.
Get Started