|
Honestly, a lot of the Hacker News discourse every single time anything having to do with Kubernetes comes up reads like uninformed annoyed griping from people who have barely or not used it. Kubernetes itself has been around since 2014. ingress-nginx was the original example of how to implement an Ingress controller. Ingress itself is not going away, which seems to a misconception of a lot of replies to your comment. A lot of tutorials use this because a lot of tutorials simply copied the Kubernetes upstream documentation's own tutorials, which used toy examples of how to do things, including ingress-nginx itself, which was meant to be a toy example of how to implement an Ingress controller. Nonetheless, it was around a full decade before they finally decided to retire it. It's not like this is something they introduced, advertised as the ideal fit for all production use cases, and then promptly changed their minds. It's been over a decade. Part of the problem here is the Kubernetes devs not really following their own advice, as annotations are supposed to be notes that don't implement functionality, but ingress-nginx allowed you to inject arbitrary configuration with them, which ended up being a terrible idea in the main use Kubernetes is really meant for, which is you're an organization running a multi-tenant platform offering application layer services to other organizations, which it is great for, but Hacker News with its "everything is either a week one startup or a solo indy dev" is blind to for whatever reason. Nonetheless, they still kept it alive for over a decade. Hacker News also has the exact wrong idea about who does and should use Kubernetes. It's not FAANGs, which operate at a scale way too big for it and do this kind of thing using in-house tech they develop themselves. Even Google doesn't use it. It's more for the Home Depots and BMWs of the world, organizations which are large-scale but not primarily software companies, running thousands if not millions of applications in different physical locations run by different local teams, but not necessarily serving planet-scale web users. They can deal with changing providers once every ten years. I would invite everyone who thinks this is unmanageable complexity to try dipping their toes into the legal and accounting worlds that Fortune 500s have to deal with. They can handle some complexity. |