A Quadrillion Rows across three Clouds: scaling LogHouse

T1Covered by 1 source · first reported by Databricks / ClickHouse / DuckDB / Supabase / Neon / PlanetScale at 30 Jun, 15:15 UTC

We scaled our internal logging platform from 19 PiB to 431 PiB and 1.59 quadrillion rows. Here’s how we rearchitected LogHouse to handle 80 GiB/s of writes while keeping queries fast and the underlying complexity invisible.

Sources

• T1A Quadrillion Rows across three Clouds: scaling LogHouseDatabricks / ClickHouse / DuckDB / Supabase / Neon / PlanetScale30 Jun, 15:15