AWS analytics feels slow by default

Hey 👋🏽

We've been running Plausible for the analytics on awsfundamentals.com. Good software, no complaints. We still wanted the data to be ours: our retention, our schema, nobody's script in the visitor's browser.

So we built the dashboard ourselves. Twice, on two different backends, fed by the same CloudFront logs.

Then we measured both instead of arguing about them.

In this issue: what AWS-native analytics actually costs in seconds and in dollars, and where it holds up fine.

SQL in, REST API out

Stream events into Tinybird, write a SQL pipe, publish it. That's your API: ClickHouse underneath, no serving layer to write, nothing to keep warm. Schemas and pipes are files in your repo, and you test a schema change on a branch of real production data before it ships.

Sponsored by Tinybird. This issue's deep dive was done in collaboration with them, and every number in it is measured from our own setup.

The honest takeaway: AWS-native isn't broken, it's just cold every single time. You can fix that yourself with a cache layer and Partition Projection, and then you own a cache layer.

See you next week! 🙏

Tobi & Sandro

AWS analytics feels slow by default

SQL in, REST API out

📚 This Week's Deep Dive