The cloud is no longer the only answer: edge computing puts processing in the last mile, close to the machines it serves.
A new idea every day
The IT industry brings a new idea to town almost every day, and most of them fade. Edge computing is one that has lasted, because it answers questions the cloud alone cannot.
Is the cloud already legacy?
First came the cloud, which promised to meet every need for processing power in exchange for a subscription. It delivered a great deal, but a plant is not an office. Some decisions cannot wait for a round trip to a distant data centre, and some data should never leave the site.
Why computing moves to the edge
Edge computing is the cloud’s sibling. It complements the cloud with computing in the very last mile, close to the devices that produce the data. The reasons are practical ones:
- Lower latency: a decision is made where it is needed, in milliseconds rather than a round trip across the internet.
- The plant keeps running: a line cannot stop because a link is down or a cloud service is unavailable.
- Less data to move: a machine produces far more raw data than is worth sending, so the edge passes on what matters.
- Control stays local: the logic that drives a valve or a motor belongs next to it, not three networks away.
- Better segmentation and privacy: the plant’s data and control stay on the plant’s own network, and the site decides what leaves.
Telecommunications carriers are looking at their exchanges as edge sites, and the idea is drawing attention across industry. Network World explains what edge computing is and why it matters.

An edge topology: computing close to the devices in the field, with the cloud behind it.
What belongs where
The edge does not replace the cloud; each does the job it suits.
The edge handles what must be fast, local or private: control, the first shaping of raw signals, alarms, and the checks that tell a real reading from a sensor that has stopped answering. The cloud handles what benefits from scale: long-term history, comparison across sites, fleet-wide analysis and the reports people read from anywhere.
The two work together through a deliberate link: the edge sends qualified data outwards, and when the link drops, the edge carries on. Deciding where each part runs is a design choice, and Control Engineering answers more questions on IIoT from cloud to edge.
CATIM’s approach
We design a blend of edge and cloud that fits each customer’s requirements, with control at the edge and the enterprise view in the cloud. CATIM Nimbus, our CODESYS edge controller with advisory AI, now in field trials, is built on that idea.
The controller is the platform. Control, data and AI run as software on one industrial unit, built on CODESYS, and the control logic is designed never to wait on the network. Sense, decide, act sets out how.
Data keeps its meaning. The controller publishes its data once, onto a Unified Namespace over MQTT, and serves the same values through the i3X API, so a new application can discover what the data means without a briefing. i3X vs MQTT explains why one moves bytes and the other moves meaning, and the controller on the Unified Namespace shows how the unit’s own health travels the same way.
Every value says whether to trust it. Each value carries its quality and the time it was true, so a device that has stopped answering never looks like a process that is simply steady. That is the silent zero, dealt with at the source.
Cloud history matches the plant. The link to an industrial cloud platform goes outbound only: nothing on the internet calls in. Each value is stored in the cloud with the time it changed in the controller, and values wait out a link loss in memory, so cloud history matches the plant.
The edge does the work, the cloud sees the whole picture, and neither has to guess what the other meant.