About fivenines.dev
Why I built it
I spent years at Uber building and running high-throughput systems, the kind that have to stay up. The most useful thing I learned there is that you do not understand a system by reading about it. You understand it once you have made decisions about it and watched what they did under load.
When I looked for that kind of learning for Kafka and Redis, I mostly found things to read and things to watch. Some were too shallow to help, some too narrow to show the whole picture, and almost none asked you to decide anything and live with the result.
That gap matters more now. AI writes boilerplate well, and it will propose a partition count, an eviction policy or a replication setup with complete confidence. Someone still has to know whether it fits what you need, and whether it will still hold next year. That takes a picture of what runs underneath, and no assistant can build it for you.
So fivenines.dev turns real systems into simulations you play. You read a service contract, build for it, run the traffic and the failures, and see what held. There is no code to write. I started with Kafka and Redis because so much is built on them, and because both are easy to start with and deep underneath.
If it helps you catch one confident, wrong proposal before it ships, it has done its job. Tell me how it goes. I read every email.
Nikolay
Who makes it
The name
Five nines is 99.999% availability: about five minutes of downtime a year. It is the bar production systems are held to, and the one every tutorial here is working toward.
The company
fivenines.dev is made by Firebrick Labs EOOD, a small software company in Bulgaria. The legal notice has its full details.
Get in touch
Write to team@firebricklabs.com with feedback, a question about your account, or just to talk systems. A person reads every email.