|
|
|
|
|
|
by datadeft
869 days ago
|
|
|
> listening ports, persistent storage, CPU, memory, ingress These exists without k8s. I do not need a compex abstraction hiding the ways I need to talk to persistent storage. If fact, I believe it is impossible to create such abstraction without very serious compromises. > In AWS, we use the ELB as our ingress, but my app doesn't care Your app does not care without k8s. Running python -m http.server does not even know what ELB is. I get it though. You are using k8s as IaC. |
|
|
That was exactly my earlier point... of course you can do everything in k8s in other ways, but in the end you have to pick ONE way your company/team is going to do it... why not pick a well defined way, that new hires can already know, that has a ton of tooling available, and works together cohesively?
Yes, I can build each part myself, but why?
> Your app does not care without k8s. Running python -m http.server does not even know what ELB is. I get it though. You are using k8s as IaC.
Sure, but I still need a way to deploy my app, and to move it to a different location when I do hardware maintenance, and a way to get a DNS address that routes to my app.
At my shop, using k8s, I can deploy a brand new service, with a cert, a url, and a place to run it, in a few minutes. I don't have to talk to anyone, I don't have to use any other tools or have to click on any buttons, i just helm install or kubectl apply and my service is running. I don't have to ask the datacenter ops people to find me a server, or get budget for a new AWS instance. I can deploy to an existing cluster and use a small bit of the infrastructure. I don't have to scale my individual service, I can scale the whole cluster for all services.
It is just so much easier to be a developer in this world.