More jabbering from me about non-clown hosting
Late yesterday, I put up a post about how to get into colocation in about the crappiest way possible. I skipped a bunch of details just to get it out there. The inspiration was based on finding out just how many people have no idea that this business model even exists.
I used to work in a cousin of this space: managed and dedicated hosting, and here's how this all lined up:
Colocation (as mentioned yesterday) gives you some space, some power, bandwidth, a network allocation of some kind (and/or ability to route your own stuff), and hopefully some decent HVAC to keep everything cool. It's your hardware in their space, and if something breaks on that stuff, you get to fix it. What if a drive fails? That's on you. Your switch or router goes insane? Same deal: that's all you.
You might be able to get some "remote hands" service from the provider for very simple tasks: reboot something, take a picture of the (blue) screen, that kind of thing. Some also let you pay them to go and do other stuff with a higher degree of complexity. Read the fine print.
Managed hosting is where you get access to a box somewhere, and are usually given root on it. You can do about what you want to it, but beware of the "spheres of support" as we used to call it - the hosting company's support people will only go so far. You might want Debian, but they only do RHEL. You get the idea. You can probably ask them to do kernel upgrades, troubleshoot why it's being slow or seems to be down, install and configure certain things for you, and so on.
If the hardware fails, that's on them. They rip open the box and replace the part. You still have to figure out how to make your stuff work before, during, and after the event. They probably won't do any migrations for you, since everyone's setup is an odd little snowflake and only the customer has any chance of knowing how it all works.
Higher-end setups might offer hardware firewalls, load balancers, backup solutions, and more. Now, granted, it's been almost 20 years since I last did this myself - I assume someone's still doing this somewhere.
Dedicated servers are a step down from managed. You (probably) get root on a box somewhere, and you can do a lot of stuff on it. Outside of the absolute bare essentials provided by the hosting company (like a Red Hat Network entitlement), they aren't going to do much for you. They might have rigged up some magic hardware to let you power-cycle what's just a boring old whitebox without explicit out-of-band access. They might also have some netboot magic to let you shunt it into a "repair environment" if you really screw things up.
All of the servers used for hosting this site were of the "dedicated" persuasion initially: first at ServerBeach, then at SoftLayer (which became IBM). The SB machines were relatively crappy whiteboxes shoved into a bread rack somewhere, while the SL/IBM stuff was on a Real Rackmounted Server with out of band remote access and the whole bit. They handled the hardware, and I handled the software.
I mentioned some of the details last year when departing the IBM shit-show and the Texas shit-show at the same time. This ended the era of dedicated hosting for me and moved me into colocation.
These days, there's a cabinet in a nice conditioned space not too far away. I initially parked an old switch in there and a derpy little Raspberry Pi, just like yesterday's post implied. This was to give me a little "platform" while the rest of the stuff happened. A few days later, I got some "real" hardware, hung that in there, and then migrated everything to it from the SL/IBM box.
That was a bit over a year ago, and things are pretty much unchanged. This setup should just sit there and run and run and run until something breaks beyond that which can be handled in software. Then I get to drive out there and flip some parts around and make it go again.
It's not rocket science. This way of doing things has always been possible. You don't have to deal with clown providers pulling the rug out from under you. There are a few dimensions in which the provider can mess with me now:
1. They could have something terrible happen - think flooding or a fire. It could take the whole location offline for an extended period, and it could well destroy the equipment inside, such that merely relocating it to another spot would not be sufficient to revive things.
2. They could decide they're done with this kind of business model, and now they're going to boot out all of the little people who pay a pittance per month. Maybe they're going to sell out to the latest clown provider who wants to melt the ice caps and drown the polar bears with their "AI" crap. (This would also apply if the private equity vultures swoop in and try to hollow the place out before nuking it entirely.)
3. They could crank the prices up beyond that which I could afford.
4. They could do something that somehow makes the basics incompatible: they stop supplying AC power, conditioned air, or IP over Ethernet. (I know, this is ridiculous, but I might as well mention it.)
5. They could shack up with some really evil people such that I no longer want to do business with them.
My responses would be approximately this:
1. Acquire some other hardware, install a fresh OS and restore my stuff from backups, then park it somewhere else.
2, 3, 4, 5. Find a new spot, then schlep the hardware there.
...
If this is old hat to you, great! It means you're probably a grizzled 1990s sysadmin just like me, consarn it! This isn't for you, then.
This is for the newer folks who might not have realized that there's an alternative to paying tribute to one of the three churches of the Clown: M, G, or A. If you want to "get your stuff online", there are other ways... and there always have been!