Posts

This project started as a coursework assignment, but the more I worked on it, the more it began to resemble a small distributed system in its own right — with all the usual problems that come with one. The Problem Weather stations are everywhere now. The cheap ones live on rooftops, balconies, and farms, and they collectively produce more environmental data than any single observatory ever did. The conventional way to turn that data into a useful rainfall predictor is to ship everything to a central server and train a model there. ...

You might be thinking: “Is this just a homelab?” And yes — technically, it is. However, running services in a self-managed environment quickly surfaces many of the same questions that appear in larger, production systems. Why? The motivation is straightforward: I needed a server that could run continuously (24/7) and host multiple long-running services, including: Jellyfin (media streaming) NAS services (storage) Aria2 (download management) Additional containerised applications running via Docker Rather than focusing on scale, the goal was to operate a real system with real constraints: limited hardware resources, a residential network, and services that should remain available without constant manual intervention. ...