Braze — Ads Analytics Pipeline Rebuild
Braze replaced a dual MongoDB + Snowflake pipeline with ClickHouse Cloud, achieving 8x performance improvement while handling 1B+ events per hour.
Scale
3.9 trillion messages and Canvas actions in 2024; over 1 billion events per hour ingested
Before
Dual pipeline — MongoDB for campaign analytics + Snowflake for raw event storage. Two systems, high operational overhead.
After
Replaced both with ClickHouse Cloud — 8x performance gain, dramatically reduced operational overhead
Key Insight
Snowflake was the incumbent — and Braze migrated away from it for real-time analytics. This is the honest tradeoff: Snowflake is excellent for batch/historical; ClickHouse wins for sub-second interactive analytics on mutable event data.
In a Snowflake Conversation
When a Snowflake prospect asks about real-time campaign analytics, this is the honest tradeoff story. Snowflake is excellent for batch/historical; ClickHouse wins for sub-second interactive analytics on mutable event data.
My Read
Practitioner commentary coming soon.
Relevant Conversations