I think you need something with a very concise value prop I'm working on that for my next thing right now. Something simple.
FYI don't use AWS for these sorts of projects use something cheaper.
I think you need something with a very concise value prop I'm working on that for my next thing right now. Something simple.
FYI don't use AWS for these sorts of projects use something cheaper.
For cost-effective startup/small scale infra, I love this method:
- get a dedicated or colocated server (can be cheaper than you think... e.g. mini PC colo with endoffice, joe's datacenter, etc.) - install microk8s on ubuntu LTS - write yaml files and deploy to microk8s
The main benefit of this is that infrastructure is very cost effective for early stage projects. Once the project needs something bigger, all of the infra is encoded in the k8s yaml files and can be deployed to a "real" k8s cluster.
Of course, the yaml needs to be carefully written and made as portable as possible so that it can be deployed to other clusters in the future.
Backup of state can be as simple as using HostPath volumes for state, and running syncthing as a deployment.
Run syncthing on other hosts (e.g. your home machine/laptop, some other cloud node, etc.) and state will be synced in realtime. Syncthing has options for keeping older versions of files/directories, so it's reasonably protected in the case the server gets hacked or deleted by accident.
This is not perfect, of course, but it's a very pragmatic and cost-effective approach. I've used/am using it and am very happy with the end result. I use cert manager, let'sencrypt, and wildcard DNS so that I can deploy new apps with their own TLS endpoints just by deploying a bit of yaml.
I used Scaleway. They don't offer the hyperscale that AWS does but you don't need that for a side project. Just get started then if you suddenly have 100,000 users then switch to something like AWS. Also there are no hidden costs it's like $2 a month for a tiny little server to get you started.
You'll find that's enough.