Tuesday, November 14, 2017

OpenShift is getting worse

As you may know RedHat provides its own public cloud infrastructure - OpenShift - for years. It was a great service, available with paid and free plans. Many companies and developers used it both for production and testing.

Unfortunately, the situation has changed since OpenShift 2 has been replaced with OpenShift 3.

First, they decided to decrease number of instances (pods, in new Openshift 3 terminology) available in the free plan from 3 to 1. I admit, this is fair. The previous limit was incredibly generous, and allowed you to run a scalable (in its minimal form, yet fully functional) cluster with a load balancer for free. That was great, but could not last forever.

Second, they restricted the uptime of the single pod to a small value, which is not sufficient for uninterrupted operation.

Third, although this is a personal one, they made the web console less convenient and intutitive. Ok, I understand they wanted to make it more flexible, but somewhat streamlined workflow similar to version 2 would be a plus as an option.

Last, but not the least. As the system became more complicated, it's now less stable and reliable. Pods can easily stuck in any state (while starting or terminating), and as a result, your site becomes inaccessible for quite a long time, because the problems can not be resolved from the console and the support does not respond quickly. And even worse: they can deliberately switch off entire clusters for "maintenance" for several days in a row.

No comments:

Post a Comment