Architecture

Lakerunner separates ingestion and query into independent, horizontally scalable paths that share only object storage and a lightweight metadata index.

• Ingestion Architecture — How telemetry flows from object storage notifications through Kafka and worker stages into optimized Parquet segments.
• Query Architecture — How queries are parsed, planned, distributed to workers, and streamed back to clients.
• Materialization Architecture — How Lakerunner precomputes selected expressions into metric segments and rewrites eligible queries to those compact paths.
• Deployment Models — Choosing between per-organization and central collector topologies.