Step 1: Tumbling Windows

This step teaches the fundamental pattern of time-series aggregation: tumbling windows. A tumbling window is a fixed-size, non-overlapping time bucket. For example, you can group all events that occurred in the same minute into a single summary event.

The Goal

Summarize each sensor's temperature readings in fixed, non-overlapping windows.

Implementation

The maintained HTTP aggregation source uses a one-minute system_window buffer. The buffer accumulates records; group_by_value separates sensors, and mappings compute statistics with from_all(). Cache expiry does not release event batches.

See the complete configuration for the Kafka variant and its combined window types.

Verification

Inspect event_count, temperature_avg, temperature_min, and temperature_max in each sensor summary. Fixture execution is distinct from testing your live source, window timing, and destination.