At the time we looked at OpenShift and decided not to use it because there was no easy wat to update an openshift cluster. In all fairness I have no clue what the current state of this issue is. For us it was important enough to be a deal breaker back then though.
We settled on Triton ( https://github.com/joyent/triton ) and while conceptually very different we are very happy with it. It allows us to leverage great technologies like ZFS, DTrace, Crossbow (network virtualization), Solaris Zones, etc.
Did you also consider OpenStack, which is what this post is about; OpenShift isn't really the same thing, and also doesn't really compete with OpnStack.
Openstack has the worst documentation of any major open source product that I’ve ever seen. It takes a team of at least three experts, who have used openstack before, to setup a cluster. Many features are completely undocumented; others are hopelessly out of date.
It’s a massive, massive project with years of technical debt. It was a bit ahead of its time, and newer projects like Kubernetes have been able to learn from its mistakes. If openstack was designed today, it would have a much different architecture.
It’s used by major enterprises, so it will be around for a while, providing jobs to the engineers who have become specialists in it.
4 comments
[ 0.16 ms ] story [ 23.6 ms ] threadWe settled on Triton ( https://github.com/joyent/triton ) and while conceptually very different we are very happy with it. It allows us to leverage great technologies like ZFS, DTrace, Crossbow (network virtualization), Solaris Zones, etc.
It’s a massive, massive project with years of technical debt. It was a bit ahead of its time, and newer projects like Kubernetes have been able to learn from its mistakes. If openstack was designed today, it would have a much different architecture.
It’s used by major enterprises, so it will be around for a while, providing jobs to the engineers who have become specialists in it.