Another midnight catches me tinkering away at my own little cloud on the Internet. Since renting these 6 machines in Hetzner I’ve been literally hacking away for the past month on setting things up fully automated (that means also dns-01 challenges for all my domains automated with Ansible + FreeIPA for the core services, DNS included and tightening up security).
Out of all 6 machines, given 6TB per machine accumulated 36TB on top of which I put a few volumes of Gluster, two of them as “backup” and “ha-vms-root-fs” as I called it. Understandably, one is for (local, fast recovery) backups, doh, one is to host HA VMs declared as “resources” in Proxmox so the cluster takes care of making them HA if one machine fails). The one for backups in addition to TLK (TurnKey Linux) which provides the “tlkbam-backup” cron.
While it’s no OpenStack or its varieties and it’s similar more to a “home-lab” (only rented or “homed” in Hetzner) it provides me with sufficient productivity and freedom at the fraction of a cost I would pay in Google, AWS or any other. It also allows me to tinker, experiment, learn.
Hetzner I would say, has hit a sweet-spot for the technology-inclined, modest-budget people and companies with ambitions. I would think that if you were a slow-growing, money constrained start-up working for the long-run, but with highly experienced folk, going the bare-metal/quick-setup servers provided here, you’d be glad. I can only see the low-cost of machines that allows small business and owners to buy a couple of them to only lend itself towards an increase in those businesses revenue and with that an increase (of resource usage) by the clients, leading to better business for Hetzner in the end.
I did a small estimation on what would it have taken to install the same services on Google for example and the price would’ve been 5x to 10x times more. Of course, much less “problems” but since I’m technically inclined, these “problems” are actually what gets up in the morning thinking and tinkering, seeing them as opportunities (to learn if nothing else).
Oh well. Time to end this. It’s late and my mind is wondering towards: “Hmmm, now that I have this small cluster almost done and with a Gluster on-top, should I put a Spark on-top and do some small data analytics experiments with the latest API just to refresh my memory?”
For another time. Meanwhile, I think I will share in the next articles some small recipes on how to get started on Hetzner + Proxmox + Gluster and how you can set-up your small “hyper-converged” cloud. Till then, enjoy.