Do I need it? No. Do I want to use it because of the superior tooling? Yes.
Honestly. Give yourself a month to learn it and migrate your home lab and it's actually not too difficult. Yes it takes a lot of debugging initially, but like most things, once you're familiar, it becomes easy.
I use Rancher for a hosted Kubernetes cluster on top of dozens of dedicated servers, and so far, it has been super nice. What are the alternatives for CI/CD for a small team (30)?
k3s is dead simple to install. Install rancher on top if you like a nice web GUI, or don't if you are fine with kubectl. The software you want to run on k8s probably has helm charts which make it easy to run.
You don't need kubernetes in the same way you don't need a fast internet service. Sure you can work with dial up, but working with 1gbps internet line is just so much more convenient
I initiated and led the Kubernetes adoption at the startup I'm with currently. Handed it off to the new ops team recently. The company is dramatically better off now in terms of scaling, deployment automation, configuration management, service discovery, internal service interactions, stability, reliability... The list goes on.
If your needs are met by a VM or three then sure, you may not need Kubernetes, although as other comments have pointed out, distributions like k3s can be useful even in those environments. But as you climb the scale and complexity ladder, there soon comes a point where it's very hard to beat Kubernetes, which is why it has become so widespread.
If you’re running containers you’ll want some kind of container management - but I’ve been fine with ECS for running on AWS. It’s well integrated, standard enough, configurable via OpenTofu/Terraform/CloudFormation etc and free for container hosts running in AWS
When I see questions like this, I wonder what the asker does instead to cobble together all that K8s does for you that you’d have to figure out. I spent years refining those ops processes. And when k8s came along and I got over the initial learning curve, it felt like I’d entered the promised land. All operational tasks that were such a pain all ended up being easier in k8s. I’ve put a small start up on k8s because of this.
9 comments
[ 3.2 ms ] story [ 261 ms ] threadHonestly. Give yourself a month to learn it and migrate your home lab and it's actually not too difficult. Yes it takes a lot of debugging initially, but like most things, once you're familiar, it becomes easy.
https://news.ycombinator.com/item?id=735568
If your needs are met by a VM or three then sure, you may not need Kubernetes, although as other comments have pointed out, distributions like k3s can be useful even in those environments. But as you climb the scale and complexity ladder, there soon comes a point where it's very hard to beat Kubernetes, which is why it has become so widespread.