Daniele Polencic
banner
danielepolencic.com
Daniele Polencic
@danielepolencic.com
270 followers 11 following 180 posts
Teaching Kubernetes at @Learnk8s.io
Posts Media Videos Starter Packs
Just landed: Learn Kubernetes weekly 154! My top picks:

⚡ Scaling Kubernetes Pain
🔐 MLOps Secret Management
🌍 DC-DR Clusters with Karmada
🧩 Kubernetes-Aware DB Proxy

Read it here: https://kube.today/issues/154
Cloud providers advertise 65,000-node clusters. 88% of teams run under 1,000

I surveyed 917 practitioners on scaling AI workloads in Kubernetes

The bottleneck isn't infrastructure, but knowing how to use what you have

Report: https://kube.today/ai-infrastructure-2025
Just landed: Learn Kubernetes weekly 153! My top picks:

🌍 Environments, Not Clusters
🔁 Terraform → Crossplane on AWS
⚙️ Local Dev Tooling for K8s
⚡ Bare-Metal Autoscaler

Read it here: https://kube.today/issues/153
🤔 Ever wondered what happens when you run kubectl apply?

This Wednesday, Rafa and I trace the complete journey through authentication, RBAC, admission controllers, and into etcd. Plus: Server-Side Apply

Join us live → ku.bz/k8s-api-ssa
🚀 Platform Engineering is coming for DevOps!

Q3 Kubernetes jobs:

- Platform roles doubled
- Software Engineers: 43%
- DevOps most remote (56%), Managers most office (59%)

Plot twist: Junior roles doubled despite AI hype

→ https://kube.careers/state-of-kubernetes-jobs-2025-q3
Just landed: Learn Kubernetes weekly 152! My top picks:

🚀 KEDA Autoscaling
💻 Cloud-Native Roles are Software Engineers
🍓 Raspberry Pi K8S Cluster Setup
🧹 Kubernetes Orphaned Resources Finder

Read it here: https://kube.today/issues/152
200 Mac minis. 20,000 iOS tests

Vitalii from Agoda explains how they brought macOS workloads into Kubernetes at scale:

- Hybrid pods
- 55-60GB VM images with COW layering
- kubectl exec over SSH

Full episode: https://ku.bz/q_JS76SvM
At KubeCon Atlanta: 4 quiz battles. 1 live leaderboard—infinite shame/glory

You will lose. You will question everything. You might win holographic stickers

Join the YAML games: https://yaml.games
Just landed: Learn Kubernetes weekly 151! My top picks:

⚙️ Kubernetes Node Stability and Performance:
📦 Dynamic PVC provisioning on AWS EKS
🎨 GPU sharing with MPS and OKE
🧹 K8s cleaner

Read it here: https://kube.today/issues/151
💣 5,000 lines of Groovy CI that only 3 people understand?

On KubeFM, Ben shared how Faire went from a Jenkins monolith to a distributed Buildkite

Key insight: Infrastructure failure rate dropped from 15% → <1%

Watch: https://ku.bz/klBmzMY5-
🎉 Launching Kube Today – the homepage for everything Kubernetes!

Daily news, jobs, events, etc., all in one place. No more jumping between sites

I've been building this since June, and it's already my default homepage

https://kube.today
Just landed: Learn Kubernetes weekly 150! My top picks:

🖥️ Production Grade K8S From Bare Metal
🔐 Kubernetes Authentication & Authorization
🐹 Deploying and Exposing Go Apps with Kubernetes Ingress

Read it here: https://learnkube.com/issues/150
"90% of teams don't need Kubernetes" - Danyl on KubeFM

Key insight: Most "microservices" are distributed monoliths that K8s won't fix. Start with modular monoliths, extract services only when you hit real constraints

Watch https://ku.bz/BYhFw8RwW
Ever had to create Helm charts from YAML files that… don't exist anymore? 🧨

I prepared a demo where Claude extracts running K8s configs, recreates the lost YAMLs, converts them to Helm charts

Next week, I will show you live: https://ku.bz/claude-k8s
Just landed: Learn Kubernetes weekly 149! My top picks:

✂️ How We Cut Our Azure Cloud Costs by 3×
📈 Resizing Pods Without the Drama
👨‍🎓 Mastering Kubernetes Security
🐍 HAMi – sharing and isolation of GPUs

Read it here: https://learnkube.com/issues/149
🧨 Adevinta's VPA went rogue, evicting critical pods in endless loops across 30 Kubernetes clusters serving 300k requests/second

The fix took seconds. Finding it took all night

https://ku.bz/rf1pbWXdN
I gave Claude root access to my server and let it migrate my K8s cluster

50+ chats: Claude provisioned K3s, secured the network, and created migration runbooks. We ran 12 dry runs until perfect

Listen to my story: https://ku.bz/claude-k8s
Just landed: Learn Kubernetes weekly 148! My top picks:

🤔 Why Kube-State-Metrics Matters
🐍 Launch Kubernetes jobs
🔒 Securing Kubernetes Resources Without a VPN
🥷 KubernetesEnumerationTool

Read it here: https://learnkube.com/issues/148
🔥 When node startup takes 9 minutes, reactive scaling fails

In the latest KubeFM episode, a student team solved this by using Facebook's Prophet to predict scaling patterns

Watch: https://ku.bz/clbDWqPYp
💡 If you want to go deeper, join me for a live discussion this Wednesday, where I will answer your GPU questions and explain how the book came to be ku.bz/g8gXCKW12
So I wrote this book. Six chapters that trace the problem from hardware to orchestration!

Download the free book here: ku.bz/gpu-k8s
This fundamental difference breaks every assumption Kubernetes makes about resources. The Linux kernel sees and controls every CPU cycle and memory page. But GPU operations? They happen in a black box managed by the NVIDIA driver. The kernel is completely blind.
Then I learned that GPUs can't even pause a running kernel. Once computation starts, it runs to completion - no preemption, no time-slicing in the CPU sense, nothing. The hardware was designed this way for maximum throughput, and no amount of software can change it.
🚀📕 The GPU + Kubernetes book is finally here. After six months of rabbit holes, I finally understood why this problem was so hard.

When I started, I thought GPUs were just fancy parallel processors. Mount the device, set some resource limits, and done.
Just landed: Learn Kubernetes weekly 147! My top picks:

🏎️ MIG on AKS
💯 Orchestrating thousands of speedtests
0️⃣ Zeropod
🔎 kubectl-sql

Read it here: https://learnkube.com/issues/147