This episode is sponsored by LearnKube - get started on your Kubernetes journey through comprehensive online, in-person or remote training https://learnkube.com/training

Gari Singh, Product Manager, Google Cloud @ Google at Google Cloud, explains how recent AI conformance and agent sandbox announcements are reshaping the Kubernetes ecosystem.

Gari emphasizes that these developments position Kubernetes as the optimal platform not just for AI training and inference, but for running agentic workloads - marking a significant evolution in how autonomous AI operations can be securely deployed and managed at scale.

Watch the interview: https://ku.bz/lwYXpLtPd

Read the announcement: https://ku.bz/s9r4CKqq5

Itiel Shwartz, Co-Founder & CTO at Komodor, shares his perspective on effective troubleshooting in Kubernetes.

He emphasizes that the key is to learn from past incidents and understand how systems worked before issues arose. Using networking problems as an example, he explains why understanding the underlying architecture and root causes is more valuable than making quick fixes to configurations without proper context.

Watch the full interview: https://ku.bz/35z_flZn3

This interview is a reaction to Alex Movergan's episode https://ku.bz/P5Y-NrSW5

Michael Levan explains how team silos in technology differ from traditional organizational silos.

He draws a parallel between specialized teams and military trenches, where teams work together in their specific domains.

The trench metaphor helps illustrate how teams can maintain deep expertise while still being part of a larger, coordinated effort.

Watch the full episode: https://ku.bz/qlZPfM-zr

Shyam Jeedigunta, Principal Engineer at Amazon Web Services (AWS), discusses two tools:

- KRO (Kubernetes Resource Orchestrator), an open-source AWS project that models resources as connected groups for deployment, enabling replication and application pattern creation.
- Karpenter, an auto-scaling and node lifecycle management tool that's gained substantial adoption and is evolving to address next-generation challenges with AI/ML workloads.
-

Watch the full interview: https://ku.bz/m89tLbgcq

Stefano Doni, CTO @ Akamas, explains what sets Akamas Insights apart from other Kubernetes optimization solutions in the market.

He highlights two key differentiators: the agent-free architecture that leverages existing observability tools instead of requiring additional cluster installations, and the comprehensive full-stack optimization approach that works across multiple levels - from cluster optimization down to pod-level application tuning and even runtime optimization for JVM and Node.js environments.

Watch the interview: https://ku.bz/rdbv-kvWt

Read the announcement: https://ku.bz/mlYTxPC6x

Stefan Roman discusses building Labs4Grabs, a Kubernetes learning platform that provides root access to users. He explains three critical security challenges:

- Implementing multi-tenancy isolation while supporting privileged workloads and root access
- Learning and configuring new isolation technologies without introducing security vulnerabilities
- Managing resource allocation to balance student experience with infrastructure constraints

The discussion explores the technical trade-offs between providing an authentic learning environment and maintaining platform security.

Watch the full episode: https://ku.bz/Xz-TrmX2F

This week on Learn Kubernetes Weekly 165:

🔥 GPU-Based Containers as a Service
🚀 Bifrost's Journey from Nginx to Envoy Gateway for Intelligent Rate Limiting
🤖 Building Production-Ready Multi-Agent Systems on Kubernetes: Deploying 11 Specialized AI Agents
🔒 Kubernetes Security Fundamentals: Networking
🐛 Debugging the One-in-a-Million Failure: Migrating Pinterest's Search to Kubernetes

Read it now: https://kube.today/issues/165

⭐️ This issue is brought to you by LearnKube — master Kubernetes with hands-on training designed for engineers who want to learn the smart way https://ku.bz/hypSbyc-V

David Pech, Staff Cloud Ops Engineer, discusses the core requirements for running Postgres on Kubernetes.

He emphasizes that teams should leverage operators with autopilot capabilities instead of building custom solutions, as database operations aren't company-specific. The platform should incorporate built-in best practices for security and configuration management, support multiple networking approaches (Kubernetes Services, PG Bouncer, pgpool), handle database extensions like PG Vector, and offer self-service capabilities through YAML for developer teams.

Watch the full episode: https://kube.fmhttps://ku.bz/rGMF2ktdb

This episode is sponsored by LearnKube - get started on your Kubernetes journey through comprehensive online, in-person or remote training https://learnkube.com/training

Ritesh Patel, Co-founder @ Nirmata, explains how their AI Platform Engineering Assistant addresses a significant gap in the market.

He discusses how AI adoption initially focused on developers for code generation, but platform engineers have been largely overlooked despite being "very stretched" and having to stay on top of many technologies.

Watch the interview: https://ku.bz/8nkrRSG_Z

Read the announcement: https://ku.bz/8_yYZZMG4

Andrew Hillier, Co-founder CTO @ Densify, discusses three key Kubernetes technologies that are reshaping resource optimization strategies. He highlights

- In-place resizing as a "holy grail" capability that eliminates the need for container restarts during resource adjustments
- AI-driven optimization with critical warnings about careful implementation rather than blindly applying AI to automation
- Karpenter for its consolidation benefits that work seamlessly with optimization tools.

Watch the full interview: https://ku.bz/-0wmZX03V

Ángel Barrera Sánchez, Senior Platform Engineer, discusses the process of setting up a managed Kubernetes cluster using GitOps principles.

The workflow involves creating the cluster via an API call, managing secrets to access private git repositories, and installing Flux for cluster synchronization.

Watch the full episode: https://kube.fm//platform-gitops-angel

Stephan Schwarz, DevOps Engineer at iits-consulting, shares how he was thrust into performance testing Kubernetes workloads.

Stephan explains why this definition phase is particularly critical for Kubernetes workloads, where the complexity of the orchestration layer adds unique considerations beyond traditional application performance testing. Rather than jumping straight into load testing tools and metrics, he advocates for establishing clear performance criteria and expectations upfront - a foundational step that shapes everything that follows in the testing process.

Watch the full episode: https://ku.bz/yY-FnmGfH

This episode is sponsored by LearnKube - get started on your Kubernetes journey through comprehensive online, in-person or remote training https://learnkube.com/training

William Morgan, CEO at Buoyant, explains how cross-zone traffic costs impact Kubernetes architecture decisions differently across cloud providers. He highlights that while Azure currently offers free cross-zone traffic (with planned future charges), Google Cloud's fast network can make latency-based load balancing less effective than in AWS. The discussion reveals how the Kubernetes ecosystem is shifting from large multi-zone clusters to smaller federated deployments, driven by these cost and performance trade-offs.

Watch the full episode: https://ku.bz/CBwn51pl-

This episode is sponsored by LearnKube - get started on your Kubernetes journey through comprehensive online, in-person or remote training https://learnkube.com/training

Brock Mowry, CTO @ Tintri, discusses KubeVirt as a key emerging Kubernetes tool he's tracking.

Coming from a virtualization background, he finds the concept of running virtualization inside containers particularly compelling. Brock has been monitoring KubeVirt's development and notes its adoption by major organizations, expressing curiosity about the project's future direction and trajectory in the cloud-native ecosystem.

Watch the full interview: https://ku.bz/F6X3C5Nvg

Yadin Porter de León, Director of Customer Marketing @ Heroku, explains how creating the "Kubernetes World: Finding Your Path" ebook addresses a fundamental gap in community onboarding.

He describes the transformation from having no structured documentation for newcomers to providing concrete resources that community leaders can hand to aspiring contributors.

Watch the interview: https://ku.bz/l97dW11Vf

Download the book: https://ku.bz/k8s-world

Dilshan discusses a real incident where migrating EKS nodes to AL2023 caused the cluster autoscaler to lose AWS permissions silently.

You will learn:

- Why AL2023 blocks pod access to instance metadata by default, breaking components that relied on node IAM roles
- How to implement IRSA correctly by configuring IAM roles, Kubernetes service accounts, and OIDC trust relationships, and why both AWS IAM and Kubernetes RBAC must be configured independently
- How to audit which pods currently rely on node roles and clean up legacy IAM permissions to reduce attack surface after migration

Watch (or listen to) it here: https://ku.bz/T_YPfTfDb

🌟 This episode is brought to you by LearnKube — join their 4-day hands-on Advanced Kubernetes course starting January 29th and finally get comfortable with production clusters. https://learnkube.com/training

With @Birthmarkb "Keep Working Harder" Farrell

Artem Lajko explains why setting up Prometheus and Grafana alone creates an incomplete observability solution.

He demonstrates how these tools only provide basic metrics without context — you might see high CPU usage in a dashboard but won't understand the root cause. This common setup fails to deliver actionable insights for troubleshooting and creates limited value for most use cases.

Watch the full episode: https://ku.bz/9sGxhmm8s