2 ms·

I think this article shows the problem with Kubernetes and how it's used and sold today. The customer needs mentioned don't align with Kubernetes and how and wh

I think this article shows the problem with Kubernetes and how it's used and sold today. The customer needs mentioned don't align with Kubernetes and how and why it is created and managed. Like the article says, people just want to run services easily, they really don't care about all the complex technical details of Kubernetes (which was born out of a highly specific technical need at Google). I think we should solve these customer needs in a simpler and more natural way. (I'm actually developing a product to do just that. Feel free to reach out to me if you feel strongly about this area).

rektide 5y ago
Here's the data in the article. > Improved maintenance, monitoring and automation - 64.6%. > Modernizing infrastructure - 46.4%. > Faster time to market - 26.5%. > Removing vendor dependencies - 12.8%. > Global reach - 12.5%. > Agility around traffic spikes - 9.2%. > Ensure portability - 8.9% Personally I don't see any misalignment. Kubernetes autonomic behavior & well established range of automation tools satisfy the overwhelming desire well. Kubernetes gets companies using a 12-factor'ish containerized delivery model, satisfying modernization. Theres a range of good Kubernetes offerings which can be brought up quickly, speeding time to market. It removes vendor dependencies. Global reach might be a miss- multi-cluster SIG is trying hard. Agility around spikes I'd chalk up as a win, as scaling clusters is easy. I don't see the misalignment. The article draws it's own conclusions. They're not well supported but I agree with them: > In other words, is a standard way of thinking about infrastructure. Going back to Richardson's earlier point, while Kubernetes and cloud-native technologies enable companies to operate at higher velocity, perhaps the biggest benefit is making skills fungible between organizations–this creates a huge performance gain for employers and employees alike. What details do you think are too complex? Nearly all of Kubernetes complexity feels reasonable to me.
- wizwit999 5y ago
  To clarify my point, I don't think Kubernetes doesn't solve those problems at all, sure it'll probably get the job done. But it's a bit like bringing a jackhammer to drill a nail. Think about it this way, if you worked backwards from those customer needs, would you really come up with the behemoth that is Kubernetes? And for complexity, sure me and you (the type of people on HackerNews) can probably handle the complexity but for many others its too much, and even I think its needlessly complex.

rektide 5y ago

Here's the data in the article. > Improved maintenance, monitoring and automation - 64.6%. > Modernizing infrastructure - 46.4%. > Faster time to market - 26.5%. > Removing vendor dependencies - 12.8%. > Global reach - 12.5%. > Agility around traffic spikes - 9.2%. > Ensure portability - 8.9% Personally I don't see any misalignment. Kubernetes autonomic behavior & well established range of automation tools satisfy the overwhelming desire well. Kubernetes gets companies using a 12-factor'ish containerized delivery model, satisfying modernization. Theres a range of good Kubernetes offerings which can be brought up quickly, speeding time to market. It removes vendor dependencies. Global reach might be a miss- multi-cluster SIG is trying hard. Agility around spikes I'd chalk up as a win, as scaling clusters is easy. I don't see the misalignment. The article draws it's own conclusions. They're not well supported but I agree with them: > In other words, is a standard way of thinking about infrastructure. Going back to Richardson's earlier point, while Kubernetes and cloud-native technologies enable companies to operate at higher velocity, perhaps the biggest benefit is making skills fungible between organizations–this creates a huge performance gain for employers and employees alike. What details do you think are too complex? Nearly all of Kubernetes complexity feels reasonable to me.

wizwit999 5y ago
To clarify my point, I don't think Kubernetes doesn't solve those problems at all, sure it'll probably get the job done. But it's a bit like bringing a jackhammer to drill a nail. Think about it this way, if you worked backwards from those customer needs, would you really come up with the behemoth that is Kubernetes? And for complexity, sure me and you (the type of people on HackerNews) can probably handle the complexity but for many others its too much, and even I think its needlessly complex.