Generic cloud machines
August 23, 2016 — July 26, 2021
Occasionally I do stuff in the cloud that is not data science in the cloud. Mostly file sync or DIY socials. For these purposes I do not require complicated orchestration of scalable Kubernetes clusters to run my high availability microservices architecture or what-have-you. I do not want backups; the service will probably be a host-proof p2p service or something else easily recreated if it falls down. What I do want is
- cheap storage, and
- cheap VMs/containers.
General considerations: Is a billion-dollar worth of server lying on the ground? tl;dr Amazon is weirdly expensive. Why not go, e.g. OVH? Is it strictly because they have fewer HOWTOs?
- Oracle Cloud Free Tier offers a lot of server for free, if you can handle it being RHEL on ARM. That sounds like a good deal to me. Have not yet tried.
- OVH is quite cheap. I use them a lot. Occasionally they catch on fire but this seems like an acceptable trade-off for my purposes, mostly about dev and experiments.
Not covered: Digital Ocean because they are too expensive.
AWS is not that cheap in list price, or in time. Last time I tried to use them I burned a whole day trying to understand their insane control panel when to upload weird XML snippets to mysterious API endpoints in order to do a relatively simple CPU-and-storage setup. The whole time I knew that if I got it wrong, the Russian mafia would steal my data.
