Why Oh Why Am I Starting a Homelab

I feel like homelab gets a bad wrap from people with more hardware on their hands than they know what to do with building overpowered systems with no real usecase. That can be cool and fun, but personally I really try to keep the things in my life that require maintenance to a minimum.

Which begs the question, why am I doing this?

Using Cloud Services

Once again, I come to you with a problem I have been thinking about recently. I launched pathfinder a few weeks ago (please play it and let me know what you think), and used fly.io to do it. It uses a not particularly resilient single-node Postgres instance, and requires a slightly cursed configuration. I have two service groups:

• one which is always on but constrained to only ever have one instance
• and another which can scale to 0 or more instances

The value of this setup is that the first service runs my cron jobs. Both services handle API traffic. This prevents having to pay the price of slow responses when the app is scaled to 0 AND having to pay for two machines being on all month. The significance of this is that with a single node for Postgres and a single node for the API server, my bill should remain under $5, which means fly.io simply won’t charge me for my usage that month, and my bill resets to $0.

This means I only have to pay if I scale up, which is really a good problem to have (i.e. it means I have USERS)! TBD on if that will happen…

Evaluating Other Hosting Options

So that got me thinking - if I want to build more prototypes or widgets or whatevers, I’m not going to be able to host them free on fly.io. Which leaves me with a few options:

1. Commit to fly.io and simply pay real money for things that will very likely have 0 ROI, effectively flushing money down the toilet,
2. Attempt to find free tier solutions on other hosting platforms, or
3. Host things on my own hardware.

Regarding option 1, I don’t feel like paying a bunch of money to fly.io to host some sites/apis that no one but me will ever use. If I ever find myself with something that has become successful and I need more stable hosting with better scaling than option 3, I would definitely consider deploying on fly.io, as it is smooth, polished, and easy to use.

For option 2, using free tier solutions is interesting but I’m not aware of options that don’t have the same pitfall as fly.io: eventually I’ll have to move on to another solution, and that means learning some new tools/processes¹. This can be cool/fun, but I’m trying to “use boring technology” more, and by doing so free myself up to focus on building whatever the core technology/value of a particular project is. So I’m not wild about option 2.

So that leaves us with option 3. Host myself. This has the obvious pitfall of having to own hardware and pay for power usage (and 1 million other little things, which is the real value of paying someone else for hosting). But I own a Raspberry Pi 3. I don’t remember why… but I have it, it’s sitting in a closet, and it can consume at most low-single-digits watts. So the financial cost is pretty negligent.

So I’m Doing the Homelab Thing

Mainly I want to host a handful of websites and services, some public (like this site) and some private. I hope to write some future posts about my explorations, so you’ll get to hear about what I get up to.

The details of my setup so far will be in my next post!

Footnotes