How Canva handles 3 million logs per second with ClickHouse
Mark Needham
Canva is a $42 billion graphic design platform with 240 million users. Their observability pipeline pushes 3 million logs and spans per second into ClickHouse. This short breakdown covers the three specific changes that made that pipeline fast and cost-effective.
- Push batching to the OTEL collector layer rather than handling it inside ClickHouse
- Design your sorting key around the queries you actually run — Canva used service name, span name, and timestamp to match Jaeger's filters
- For event-driven log ingestion (Kinesis + Lambda), async inserts eliminate blocking and guarantee reliability
- The payoff: 14x compression ratio, 70% storage cost savings, and trace searches that now complete in 2.5 seconds instead of 30
Open House
Scaling ClickHouse to petabytes of logs at OpenAI
Open House
How ClickHouse helps Anthropic scale observability
Open House, User stories
How Capital One cut infrastructure costs by 50%
Engineering leaders at Capital One share how they cut infrastructure costs by 50% and reduced average dashboard load time from 5+ to under 500ms with ClickHouse Cloud.
