Revisiting CentOS Atomic Host
A couple years ago, I wrote an article about how I was choosing CoreOS over Project Atomic based on some initial testing with CentOS Atomic Host builds. As it turns out—and as I pointed out in the “Update” section of that article—the Atomic Host builds I was using were pre-release builds, and therefore it wasn’t really appropriate to form an assessment based on pre-release builds. Now that both CentOS Atomic Host and CoreOS Container Linux have both grown and matured, I thought I’d revisit the topic and see how—if at all—things have changed.
In my original post, there were 4 major issues I identified (not necessarily in the same order as the original post):
- Lack of container-specific cloud-init extensions
- Difficulty customizing Docker daemon
- Odd issues with cloud-init
- Stability of the distribution
So how do these areas look now, 2 years later?
-
Container-specific cloud-init extensions: Upon a closer examination of this issue, I realized that the cloud-init extensions were actually specific to CoreOS projects, like etcd and fleet. Thus, it wouldn’t make sense for these sorts of cloud-init extensions to exist on Atomic Hosts. What would make sense would be extensions that help configure Atomic Host-specific functionality, though (to be honest) Continue reading
Twice as many CEOs are building up in-house IT rather than outsourcing.
It fills a gap in IoT related to the last mile of networks.