Case Study: Netflix optimizes petabyte-scale logging with ClickHouse

How Netflix optimized its petabyte-scale logging system with ClickHouse

Netflix needed a way to make its petabyte-scale logging system searchable and interactive at extreme volume. The company ingests about 5 petabytes of logs per day across more than 40,000 microservices, with millions of events per second and hundreds of queries per second, so it needed a fast hot tier for recent data. Netflix used ClickHouse as the core of that hot storage layer, alongside Iceberg for long-term retention.

ClickHouse helped Netflix scale by supporting three key optimizations: a generated lexer for log fingerprinting, a custom native-protocol serializer for high-throughput ingestion, and sharded tag maps for faster queries. These changes boosted fingerprinting throughput by 8–10x, cut average fingerprinting latency from 216 to 23 microseconds, and reduced common tag-filter queries from about 3 seconds to 1.3 seconds or under 700 milliseconds. With ClickHouse, Netflix now makes logs searchable within roughly 20 seconds, sometimes as fast as 2 seconds for live streaming.

Netflix

Daniel Muino

Engineer