Also, holy FUCKING SHIT. DEPLOYING A KUBERNETES CLUSTER JENKINS AND SONARQUBE EC2s, TAKES SO MUCH FUCKING RESOURCES AND COSTS SO FUCKING MUCH THAT THIS HURTS ME FUCKING PHYSICALLY.

HOW DO I PRACTICE WORKING DEVOPS WITH KUBERNETES ON AWS IF THE COSTS ARE FUCKING OUTRAGEOUS?????

You're better off to rent a server and manage services just with systemd and docker.

get a beefy physical server at Hetzner or whereever and use it as your k8s worker node. This way you'll have decent performance out of a single server for a fixed price of ~50€/mo.

Unless you're focusing on managed k8s and not k8s in general...

My whole infrastructure is just apps on raw Ubuntu (no docker) behind nginx, orchestrated by simple bash scripts written by ChatGPT. Works flawlessly so far, one year and counting

Get an account on DigitalOcean,

spin up a decent Droplet and put everything there.

No need to use AWS managed service, they're expensive AF.

Would bankrupt a startup for sure.

As others have said, get a big vm or one+ physical computer, and install k8s.

K3s is quite easy to set up and get started.

SaaS kubernetes is a subset of kubernetes "from scratch" so if u learn to manage k3s you will have all the tools you need pretty much. In fact its a bit overkill.
But on prem k8s stuff could be a good skill to have.

If you do go for k3s, and you want smth more that bare kubectl, you can use Rancher as a management UI.
Terraform also supports it, if you wanna go down that path.

There is surprisingly little that is unique to AWS/Azure, its all the same stuff with different names

@Elendil
PS: i have a visceral, deep distaste of using tools that only work on one specific cloud vendor. If you literally only care about Jeffs walled garden, maybe there are better ways of acheiving your goals.

You are using cloud wrong. Virtual machines are meant for lift and shift scenarios.