Over the years, many graphical user interfaces have appeared for Kubernetes, but only a few continue to be actively maintained and developed. While I prefer the standard kubectl CLI for basic interactions with a Kubernetes (k8s) cluster, a visual representation can be extremely helpful for troubleshooting and debugging.
Headlamp was my go to UI tool for k8s, which provides a simliar user experience to many of the previous Kubernetes visualization tools, nothing fancy but it gets the job done.
I recently discovered another OSS Kubernetes visualization tool called Radar, which aims to fill some of the gaps left by existing tools on the market. I was a bit skeptical as this could just be another "modern" tool until the next one is built, but I figured I would give it a try!
At first glance, it has a much cleaner, modern look and feel after I connected it to my vSphere Kubernetes Service (VKS) cluster. The homepage provides a summary of all the expected datapoints including versioning, resource utilization, deployment details, and the health of components and workloads.
Like many other k8s UI tools, after installing Radar, it will use your current k8s context or you can also manually export the KUBECONFIG environment variable. To launch Radar, simply run the following command:
kubectl radar
Note: For Windows users, you will use the kubectl-radar command.
Once running, Radar will automatically open your default web browser to http://localhost:9280
Maybe it is just me, but I prefer having a flat navigation experience, especially within the topology view compared to other tools which typically requires multiple clicks, expanded menus and then selecting or de-selecting the resources that you want rendered on the screen.
It feels much more responsive and more importantly, allowing you to quickly drill down into the thing you want to look at.
In the upper right corner, you can scope the output based on a specific k8s namespace
I also really apperciate the clear visualization of the CPU and memory utilization which also includes rendering both requests and limits, which can be difficult to grok when you are troubleshooting resource issues.
The timeline is a pretty unique feature that can really come in handy when troubleshooting a complex k8s environment, giving you a visual representation of a deployment and seeing where something may have failed or degraded. This is definitely something I plan to explore further with some of my workloads that I am working with 😅
If you are a VMware Cloud Foundation (VCF) user, you can also connect up your VCF Automation (VCFA) which is a native k8s application to Radar as shown in the screenshot below.
Here is the topology view for a more complex k8s namespaces:
Here is an example of a running Pod and its resource utilization that shows not only the current demand but also the configured requests and limits in a super intuitive visual.
I really like what the Skyhook OSS team has done with Radar, I am definitely a fan and will be using it as my go to k8s visualization!
Thanks for the comment!