7 Comments

Great benchmark!

Made me think about recent discussions I had with Kestra users which are using Kubernetes + DuckDB (and a bit of Kestra) to manage such problem.

Not the serverless way indeed, but same idea of using DuckDB as the query engine for low latency requests ✅

Expand full comment

Hum, the duckdb + lambda benchmark seems to very slow with python. From my (limited) experience using Rust with Duckdb is quite fast and a lot cheaper.

Expand full comment

yes good point, havn't try deploying duckdb with rust.

Expand full comment

Loved this post. Would be awesome if you did more cost / performance comparisons with ETL tools and data transformation / integration libraries for pipelines ( dlthub, polars, airbyte ) for example.

Expand full comment

Thanks Ibrahim. yes indeed, great idea for a future post :=

Expand full comment