VPS vs in-house servers

Everything I talk about on this blog runs on real machines: an internal network of 8 servers and an external network of 6 VPS, which are servers rented on the internet. In this section I explain why I have both, what runs on each and how I decide whether something goes inside or outside: the data, the cost, the availability and how exposed it ends up.

Everything to do with infrastructure belongs here: each machine's hardware, the operating system, how I deploy and configure, the network between home and outside, and what it all costs to maintain. The practical side: how I set up a new VPS before putting it to work, what I do so it doesn't get knocked over in the first few hours, when I use Linux and when Windows, and where each agent, harness and model lives. How models are chosen and used I cover in Local AI. And all of it with the mistakes I made, because most of what I know I learned fixing something I'd done wrong.

And the usual question: is it worth it? I give you the numbers for what the VPS and the in-house servers cost me each month, in money and in maintenance hours, and who each option makes sense for: a freelancer, a small business or someone who enjoys tinkering. You don't need fourteen machines to get something useful out of this section.