Use clickhouse instead of Mongo as backend storage? about growthbook

[jdorn]

MongoDB is only used to store metadata and cached query results. The actual raw experiment and metric data can be stored in Redshift, BigQuery, Snowflake, or a few others. I think it definitely makes sense to add ClickHouse to that list of supported data sources.

I'm not too familiar with ClickHouse. Are there any major limitations compared to the other big columnar databases that might make it difficult to support? We don't use window functions, so that makes things easier I think.

[Slach]

clickhouse even support p-value calculation https://clickhouse.tech/docs/en/sql-reference/aggregate-functions/reference/studentttest/