Case study
nodela
Phase 1A domain-agnostic IoT backbone
Turns raw field packets into canonical events, runs your stateful rules on top of the stream, and writes the result to a live map and to alerts. Installs into your own cluster with a single Helm chart.
- ≤ 5 s
- save a rule, the live pipeline takes over
- 3 codecs
- JSON · CBOR · MessagePack, one canonical event
- One chart
- broker to dashboard, whole stack
Why it started
Teams with devices in the field hit the same wall: every device type speaks differently, every customer wants different rules, and two customers later you maintain three copies of one system. Off-the-shelf IoT hubs solve that repetition but arrive with licence costs and whitelabelling limits that make it hard for a partner to sell under its own brand. nodela sits between the two: a backbone that removes the repetition while running inside the organisation's own cluster, under its own brand.
Key decisions
- 01
A device type should be a record, not code
Supporting a new device means writing a manifest and a JSON schema: fields, IO map, field aliases. Two devices reporting the same reading under different names collapse into one canonical event. Diversity in the field no longer requires a release — that was the thing actually slowing the product down.
- 02
Rules on the hot path, with a drawn boundary
Rules are written in TypeScript and the live pipeline takes over the moment they are saved; no redeploy. In exchange, long-running workflows were deliberately kept off this path. That is a boundary drawn to keep hot-path latency predictable, not a missing feature.
- 03
A packet that fails validation is never dropped silently
Data that does not match the schema goes to the failure log instead of disappearing. Real device output is the only reliable way to learn a protocol in the field, and a lost packet is an error nobody can learn from.
- 04
Not SaaS — one chart
The whole stack, from broker to dashboard, installs into the organisation's own cluster with a single Helm chart, and the data stays inside. In enterprise purchasing we found this weighs heavier than price.
Where it is today
Phase 1 is up: the end-to-end path runs from MQTT to canonical events, from the rule engine to a live dashboard. Whitelabelling and multi-tenancy are defined in the first schema.
The product itself
The feature list, screenshots and documentation live on the product's own site.