The auto-repair function in Kubernetes is perfect. When something breaks, the auto-repair function automatically repairs it. If you are running the content in Kubernetes, you have a good set up. You do not need to do anything for the management of this. So, the automation of Kubernetes is number one.
The autoscaling feature is the most valuable. Kubernetes itself is an orchestration tool. It automatically detects the load, and it automatically spins up the new Pod in the form of a new microservice deployment.
Azure DevOps and Cloud Lead at a consultancy with self employed
Sep 27, 2022
The Desired State Configuration is a handy feature; we can deploy a certain number of pods, and the tool will ensure that the state is maintained in our desired configuration.
There are many good features. I feel that the scale-out features, like replica sets, are very good. The number of running containers can be autoscaled.
In the financial service sector, I'd rate scalability an eight out of ten. But do it in a controlled manner, not auto-scaling. If your application has a bug and you enable the autoscaler, it will spike your costs. If someone deploys an application with a bug, that's automatically a problem.
The Kubernetes dashboard can be improved. It is currently a mess. We were using Rancher earlier, and everyone was happy with the dashboard. Right now, we are using Kubernetes, and it's not working with Microsoft workstations. We still have problems with the dashboard. It's terrible.
I'm expecting more improvement on the UI development side, which can be reflected in each object that is part of Kubernetes, like the Pod, deployment set, ReplicaSet, ConfigMap, Secrets, and PersistentVolume.
It would be very interesting if they could introduce a template engine to set dynamic values in the deployment time. It would be ideal if it could be native in Kubernetes as it would be much easier.
Azure DevOps and Cloud Lead at a consultancy with self employed
Sep 27, 2022
The solution has some issues regarding availability during high loads. Worker nodes are sometimes unavailable, affecting the overall availability of the applications. This is a bug or underlying problem with the tool, and Azure and other providers are looking into improving this by releasing new versions of Kubernetes that fix some of the platform's issues.
Multi-Cloud Consulting at a construction company with 5,001-10,000 employees
Jan 30, 2021
One feature I would actually like to see is the network monitoring part. When we talk about communities, it's mostly the computer side. But it does have some enhancements on the networking side which they have recently released. I would like to see more enhancement where we can monitor the networks of the Kubernetes cluster or the Kubernetes workloads.
It's good for bigger organizations, but for smaller organizations or a few workloads, it may be too heavy, not easy to deploy, and the ROI may be less because it requires a control plane, worker nodes, and multiple VMs to run.