3 ms·
I'm surprised no comments here mention containers. These are a great example of where IBM absolutely screwed itself. Most of the kernel features for containers
by contingencies 11y ago
I'm surprised no comments here mention containers. These are a great example of where IBM absolutely screwed itself. Most of the kernel features for containers and the initial release of LXC were done on IBM's dime. The two major developers (one focused on kernelspace, one on userspace) then left, at least one of them for Ubuntu, I believe. I know this because I was in touch with them and using LXC personally about 2009-2010 before the docker hype era.
As I comprehend it, the features were originally intended to allow fat mainframes to dole out tiny resource portions for distinct jobs. As it seems to have turned out, every CPU these days is the equivalent of a fat mainframe, and the greater challenge is around integrating this form of rapid SOA micro-provisioning in to developer workflow through tooling and education. I just upgraded this few year old Macbook Pro to run Docker 1.1.0: 200MB Linux ISO + Virtualbox + kludgey go wrapper = better solution than IBM ever provided. With a different management culture, they could have released this in 2010, pre-empting stable docker by 5+ years with a kernel commit history and adequate marketing funding.