Spotlight

What Does 4.4% GPU Utilization Actually Mean?

Federico Iezzi

This article explains why low GPU utilization during LLM inference can be normal by breaking down prefill, decode, memory bandwidth limits, batching, and GKE B200 benchmark numbers.

More articles →

Tools and utilities

  • Kube-Argus

    Kube-Argus is a single-binary Kubernetes dashboard that combines live cluster state, log streaming, YAML editing, drain workflows, cost analysis, and AI-assisted diagnosis in one web interface.

  • Kaniop: Kubernetes Operator for Kanidm

    Kaniop is a Kubernetes operator written in Rust for managing Kanidm identity management clusters, providing declarative identity management through GitOps workflows.

  • Kubetail: real-time Kubernetes log dashboard

    Kubetail is a Kubernetes logging tool that streams workload logs into a browser or terminal, merges multi-container logs into one timeline, and works without sending logs to an external service.

  • Xata: open-source Postgres platform on Kubernetes

    Xata is a Kubernetes-native Postgres platform for running many databases with copy-on-write branching, scale-to-zero, autoscaling, HA, backups, REST APIs, CLI, and RBAC.

  • KSail: Kubernetes SDK

    KSail is a unified Kubernetes SDK that simplifies cluster management and workload deployment by providing a single, declarative interface for all your Kubernetes operations.

More projects →

Events starting soon

Discover more events onn Kube Events →

The Hidden Cost of Slow Autoscaling
The Hidden Cost of Slow Autoscaling

Forced platform migrations are usually treated as something to survive. At Scout24, a mandatory OS migration became an opportunity to rethink Kubernetes autoscaling, node provisioning, and infrastructure efficiency.

John Ford explains how Scout24 moved its EKS-based Infinity platform from a polling autoscaler and over-provisioned capacity to Karpenter and Bottlerocket. The result was faster node startup, a safer migration path, and about a 30% infrastructure reduction without major downtime.

In this interview:

  • Why two-minute node provisioning forced a 25% capacity buffer
  • How Karpenter made the Bottlerocket migration safer
  • What broke around EC2 metadata, AWS SDKs, and cgroups
  • How the new foundation enables Spot, ARM, and GPU workloads

Learn from production

More case studies →

Matching jobs

    • Data Engineer with Foundry Robotics

    • Salary: $150K to $250K a year

    • Location: based in the office in San Francisco, CA, USA

    • Tech stack: Kubernetes, AWS, Azure, GCP, On-premise, Go, Java, Python, SQL, Typescript

    • Data Engineer with Neura Robotics

    • Salary: US$88.2K to US$247.5K a year

    • Location: based in the office in Metzingen, DE; Riederich, DE

    • Tech stack: Kubernetes, AWS, Azure, GCP, Docker, Java, Python, Scala, SQL, Kafka

    • DevOps Engineer with Career Team

    • Salary: $81K to $297K a year

    • Location: remote from

    • Tech stack: Kubernetes, AWS, Flux, Docker, Javascript, Python, Typescript, Cloudformation, Terraform, GitHub Actions

    • DevOps Engineer with Egen

    • Salary: $117K to $261.91K a year

    • Location: fully remote

    • Tech stack: Kubernetes, GCP, On-premise, Terraform

    • DevOps Engineer with Netcompany

    • Salary: US$121.5K to US$275K a year

    • Location: based in the office in The Hague, NL

    • Tech stack: Kubernetes, Docker, Powershell, Terraform, Azure DevOps, Ansible, Grafana, Prometheus, Loki

Discover more Kubernetes jobs on Kube Careers →

Subscribe to Learn Kubernetes Weekly

Trusted by 77K engineers. Delivered 190 issues and counting.

or subscribe via

Build something

More tutorials →

Call for Papers closing soon

  1. 2

    days

    AI Coding Summit

    The Call For Paper is open until 9 July 2026 at GMT-4. More info →
    • Location: New York, NY, USA

    • In-person conference organized by GitNation.

    • The conference starts on the 19 November 2026.

    • Apply here
  2. 7

    days

    AI Engineering Summit

    The Call For Paper is open until 14 July 2026 at GMT-4. More info →
    • Location: Berlin, DE

    • In-person conference organized by AI Engineering Summit.

    • The conference starts on the 20 November 2026.

    • Apply here
  3. 7

    days

    Kubernetes Community Days Gujarat 2026

    The Call For Paper is open until 14 July 2026 at GMT-4. More info →
    • Location: Ahmedabad, IN

    • In-person conference organized by KCD Gujarat.

    • The conference starts on the 19 September 2026.

    • Apply here
  4. 8

    days

    Kubernetes Community Days Suisse Romande 2026

    The Call For Paper is open until 15 July 2026 at GMT-4. More info →
    • Location: Geneva, CH

    • In-person conference organized by KCD Suisse Romande.

    • The conference starts on the 10 December 2026.

    • Apply here
  5. 8

    days

    Great International AI Native Summit

    The Call For Paper is open until 15 July 2026 at GMT-4. More info →
    • Location: Bangalore, IN

    • In-person conference organized by Saltmarch.

    • The conference starts on the 10 December 2026.

    • Apply here
  6. 9

    days

    Kubernetes Community Days Porto 2026

    The Call For Paper is open until 16 July 2026 at GMT-4. More info →
    • Location: Porto, PT

    • In-person conference organized by KCD Porto.

    • The conference starts on the 20 November 2026.

    • Apply here
  7. 10

    days

    Subsurface 2026 — The Open Data and AI Conference

    The Call For Paper is open until 17 July 2026 at GMT-4. More info →
    • This is a virtual event

    • Online conference organized by Dremio.

    • The conference starts on the 2 October 2026.

    • Apply here

Thanks to our sponsors who make Kube Today possible

Find out more about being a sponsor →

More articles

Even more articles →