# Foundations

The handful of concepts you need to build with Node-RED, and how they fit together. You wire pre-built nodes into flows and spend your effort on the logic — what the system should do — while the platform handles the syntax and the connectivity. Learn these and you can build real integrations in Node-RED with only a little JavaScript — the structure carries most of the weight.

## How it fits together

<flow-diagram :edges="["msgin>transform","transform>route","route>sink",{"from":"route","to":"context","dir":"both","accent":"green","label":"read / write state"}]" :nodes="[{"id":"msgin","label":"Message in","sub":"http in / inject"},{"id":"transform","label":"Node","sub":"transform"},{"id":"route","label":"Node","sub":"route"},{"id":"sink","label":"Sink","sub":"no routing"},{"id":"context","label":"Context","sub":"shared state","accent":"green","col":3,"row":2}]">



</flow-diagram>

A **message** enters a **flow** and passes from **node** to node, transformed along the way, until it reaches a sink that sends and routes nothing. Reuse comes from **subflows** and **link nodes**; shared state lives in **context**; new capabilities come from installing nodes off the **palette**.

## What it connects to

Node-RED's reach comes from its nodes: install one for a protocol or service and the instance can talk to it. The same runtime — a Hosted Instance in the cloud or a Remote Instance on your own hardware — reaches field hardware, databases, message buses, cloud services and other systems. And **any instance can serve its own Dashboard** for the people who use it.

<arch-diagram :edges="[{"from":"inst","to":"dash","label":"serves"},{"from":"inst","to":"db","dir":"both"},{"from":"inst","to":"mqtt","dir":"both"},{"from":"inst","to":"cloud","dir":"both"},{"from":"inst","to":"plc","dir":"both"},{"from":"inst","to":"io","dir":"both"},{"from":"inst","to":"gw","dir":"both"},{"from":"inst","to":"api","dir":"both"}]" :nodes="[{"id":"dash","label":"Dashboard","sub":"its own operator UI","accent":"blue","col":1,"row":1},{"id":"db","label":"Databases","sub":"SQL · time-series","accent":"green","col":2,"row":1},{"id":"mqtt","label":"Brokers","sub":"MQTT · UNS","accent":"teal","col":3,"row":1},{"id":"cloud","label":"Cloud services","sub":"AWS · Azure · GCP","accent":"blue","col":4,"row":1},{"id":"inst","label":"Instance","sub":"hosted or remote — same runtime","accent":"indigo","span":2,"col":2,"row":2},{"id":"plc","label":"PLCs","sub":"controllers","col":1,"row":3},{"id":"io","label":"Sensors & IO","sub":"signals","col":2,"row":3},{"id":"gw","label":"Gateways","sub":"protocol bridges","col":3,"row":3},{"id":"api","label":"HTTP / APIs","sub":"REST · services","accent":"blue","col":4,"row":3}]">



</arch-diagram>

## The core pieces

- **Node** — A single processing block: it receives a message, does one thing — read, transform, call, route — and passes it on.
- **Flow** — Nodes wired left-to-right on a tab; a message enters, is transformed, and exits. One working unit of automation.
- **Message (msg)** — The object that travels the wires, carrying **msg.payload** plus metadata between nodes.
- **Subflow** — A block you define once and drop in many places, with its own inputs, outputs and per-instance config.
- **Link nodes** — link in / link out route messages across tabs with no visible wires; **link call** is the one that *returns* — your in-process service call.
- **Context** — Storage that keeps state between messages — flow and global scope, in memory or persisted.
- **Palette** — The library of installable nodes (npm) you add new capabilities from.
- **Editor & runtime** — The browser editor where you wire flows, and the runtime that executes them continuously.

<callout icon="i-lucide-book-open">

**In the Node-RED docs** — that's the working model. For the full glossary — every core term and node type, straight from the Node-RED project — see the [official Node-RED documentation](https://nodered.org/docs/user-guide/concepts) instead of a glossary here.

</callout>