Connect your data

Frames read from tables in ClickHouse. Before a dashboard shows anything, two things have to be true: Noemata is pointed at a database, and that database has data. This section covers both — where the rows live, and how they get there.

The default local stack wires all of it up for you. noemata up starts a managed ClickHouse, starts an OpenTelemetry collector in front of it, and installs the semantic layer and dashboards for whatever it detects. The pages here are for when you want to change that — an external ClickHouse, a custom collector pipeline, or a different set of integrations.

The pieces

• A database. ClickHouse holds every table a frame can query — your telemetry plus Noemata’s own file store. It’s a managed local binary by default, or an external server / ClickHouse Cloud you point at.
• A collector. The bundled OpenTelemetry collector receives OTLP and writes it to ClickHouse. Integrations configure it to pull from specific sources (log files, database stats, host metrics).
• Integrations & packs. An integration detects software you run, sets up its collection, and installs a frame pack — a ready-made semantic layer and dashboards for it.
• Editor skills. The noemata integration writes skills into your coding agents so they know how to author frames in the workspace.

For how these fit together, see Architecture.

In this section

• Connect a database — ClickHouse, local or external.
• Send telemetry — the OpenTelemetry collector and OTLP.
• Integrations & packs — semantic layers and dashboards you install.
• Editor skills — teach your coding agent the workspace.

If you haven’t run the stack yet, start with Getting started — the setup wizard makes the common choices for you, and everything here is how to override them afterward.