Kubernetes has given many of our clients, especially in finance, more control and flexibility over how they run their software. But when we looked closely at some of their clusters, the same problems kept coming up: poor visibility, resources that weren’t tuned properly, and costs creeping up as a result.

The problem with flat dashboards

Tools like Grafana and the Kubernetes Dashboard show metrics in 2D, which works well until a cluster gets large. At scale, it’s hard to see which workloads have too much or too little CPU and memory. Teams often over-provision as a safety net, which means paying for capacity they never use, and a bigger carbon footprint.

Stepping inside the cluster

We built a proof of concept that brings Kubernetes monitoring into virtual reality. Using Unity, an Oculus Quest headset and the Kubernetes API, users can walk through a 3D model of their cluster and see pods, nodes, memory use and CPU load in real time.

Height, colour and position do the job a table of numbers usually does, so it’s obvious at a glance where resources are under-used or over-allocated.

What we found

• The prototype surfaced resource inefficiencies in a live cluster running more than 870 pods.

• Clients used it to reallocate CPU and memory more effectively.

• Testers spotted performance issues faster than they could in a traditional dashboard.

“By bringing Kubernetes into VR, we’re making infrastructure data more accessible, more intuitive and more actionable.”

What’s next

There’s still work to do on real-time metrics, the interface and performance at scale. Next, we’re exploring:

• Streaming metrics in real time

• Richer visual metaphors

• Multi-user collaboration in VR

• Cloud API integration for predictive insights

If you’d like to see how immersive technology could help your DevOps team, we’d love to hear from you.