Case study
HearthCraft was a free, community-driven Minecraft server I led from 2018 to 2025. It started as a small world for friends and grew into a live community with staff, player support, hosting bills, and a strong no-pay-to-win rule.
Reasoning
A managed cloud stack would have cut some admin work, but the monthly bill climbed fast. A dedicated bare-metal box from Bloom.host gave predictable CPU and RAM for one flat cost that donations could cover. Paid plugins could have bought convenience, but they would have pulled the economy toward unfair advantages.
Choice
I bought that dedicated Bloom.host machine, containerized the Java server in Docker, and kept restart and backup scripts simple enough for volunteer staff to run. The shop funded hosting without selling gameplay power.
What I'd repeat
The work was not only keeping a process alive. I treated lag spikes, grief reports, and staff mistakes as production incidents, with backups, rollback habits, and runbooks instead of panic fixes.
Outcome
HearthCraft ran for seven years without resetting the world. Players logged thousands of hours, donations covered hosting, and the project became my first serious lesson in operating software for people who relied on it daily.
Postmortem: If I were rebuilding it now, I would add better telemetry and migration scripts so staff could spot performance and data issues earlier.
Impact: 50,000+ players · 7 years · 0 world resets · donation-funded ops
