The tiny open-source tool that replaced a $40k logging stack
One engineer's weekend project is now running in production at three banks — and it fits in a single binary.
It started, as these things often do, with a bill. A mid-sized fintech was paying more than $40,000 a year to ingest and search its logs — and most of that spend went to data nobody ever looked at.
So one of its engineers spent a weekend writing a stripped-down alternative: a single binary that tails logs, indexes only what matters, and answers the handful of queries the team actually runs during an incident.
Boring on purpose
The tool does far less than the platform it replaced, and that is the point. There is no query language to learn, no cluster to babysit, and no per-gigabyte meter running in the background.
“Most observability spend is insurance against questions you never ask. We decided to stop paying the premium.”
Eighteen months later it is open source, running in production at three banks, and quietly proving that a lot of expensive infrastructure exists mostly because nobody questioned it.
