Skip to content

Data pipelines

Moving data from where it lands to where analysts and models can reach it.

Classic ETL

Simplest possible pipeline, shown in its simplest form. Sequential, uniform node size, one theme colour — intentionally flat so the following examples read as elaborations.

Streaming with consumers

Kafka-style event log fanning out to three consumers. The log node takes visual weight via sizeScale; a purple theme signals "stream domain."

Lambda architecture (speed + batch)

Two lanes over the same events. Regions separate the real-time and scheduled bands visually; annotations name the latencies.

CDC → lake → catalog

Change data capture into a lake, discoverable via a catalog. Badges mark the "health" of each CDC feed.

Edit this pageLast updated: