About
I'm an engineer working on infrastructure and data systems: Kubernetes in production, PostgreSQL and ClickHouse for sensor data at scale, and the tooling that keeps it all observable and affordable.
This site is where I write things down. Most posts start as a question I couldn't find a measured answer to — how do Hetzner's storage classes actually behave under database load? Is a Kafka-less CDC pipeline into ClickHouse viable? — and end as a benchmark with reproducible numbers. I try to state the limitations as clearly as the results.
I'm not selling anything here. If a post saves you a week of evaluation, it did its job.