What Is Edge Software?
The software that makes local physical computing connectable, observable, and operable.
Edge software is the local software layer that connects physical sources to useful operational outcomes. It may acquire device data, run local tasks and models, store evidence, evaluate events, expose dashboards, monitor runtime health, and deliver selected results to other systems. It is more than a device driver and less than a replacement for every enterprise application.
Give local hardware a repeatable operating role
A computer near a physical process becomes an edge node when its software can reliably perform a defined local role. That role can include device connectivity, data normalisation, local processing, evidence retention, event creation, security controls, health monitoring, and integration. The software turns general-purpose hardware into a supportable part of a wider physical operation.
Edge software is not one fixed product category. A deployment can combine operating-system services, protocol adapters, a device gateway, custom code, a model runtime, a local database, an event engine, and a remote-management layer. The useful architecture is defined by responsibility boundaries, not by the number of packages installed.
Understand the layers before choosing a platform
| Layer | Purpose | Typical examples |
|---|---|---|
| Host and runtime | Run local processes, storage, networking, and update procedures. | Windows, Linux, container runtime, service manager, local filesystem. |
| Device and protocol integration | Connect field equipment and normalise source state. | Camera, USB, Modbus, MQTT, serial, file, API, or vendor-gateway integration. |
| Local processing | Transform, analyse, classify, or score local inputs. | Rules, signal processing, image or AI inference, Python/native service, custom module. |
| Orchestration and response | Turn results into events, retained records, alerts, actions, and controlled outputs. | Event engine, action runner, workflow, dashboard, local history, notification service. |
| Operations and integration | Make the node supportable and connect selected outcomes to authorised consumers. | Health monitor, access control, remote diagnostics, MQTT/API/file/database handoff. |
Do not ask one edge layer to replace every specialised system
Edge software can surround a specialist controller, model, robot runtime, BMS, SCADA, WMS, or enterprise application with local connectivity and operational context. It does not automatically replace that system’s domain authority. A PLC or safety controller remains responsible for its designated control function; a model remains responsible for its analytical method; a system of record remains responsible for the business state it owns.
The integration boundary must make this visible. Define which inputs enter the edge layer, which results leave it, who owns each schema, and what happens during failure. An explicit boundary is easier to test, secure, replace, and explain than an undocumented direct connection between every device and application.
Local components, tasks, events, and controlled delivery
Banalytics packages edge software into a local Agent and independently versioned modules. Components represent connected resources; tasks perform local work; Event Manager evaluates structured events; actions execute preconfigured responses; dashboards and history expose the operational result. This lets a site add the capabilities it needs without asking a custom model or device driver to become the entire operating platform.
For example, Modbus Line can poll and track local device state, the embedded MQTT Server can provide a local publish/subscribe boundary, and Event Manager can route a defined result to an operator or external integration. These are not generic claims about every protocol; they are explicit local capabilities with their own configuration and operational contracts.
Choose edge software for the operational work around devices
- It must connect the actual sources at the site or support a controlled gateway for the ones it does not connect natively.
- It must retain source time, data quality, error state, and the evidence required by the operating workflow.
- It must support updates, credentials, access roles, health monitoring, storage management, and recovery.
- It must deliver selected outcomes to other systems without forcing every raw input across the same boundary.
- It must preserve safety, control, and system-of-record authority where those responsibilities already exist.
Good edge software makes physical computing operationally ordinary: local enough to be resilient, structured enough to be supportable, and open enough to connect to the systems that need its results.