Kubernetes
Kubernetes Wasn’t Built for GPUs. Make It Behave
Kubernetes counts whole GPUs and treats pods as disposable. An LLM pod is neither. Share the silicon with MIG/MPS/time-slicing and stop paying for idle ...
Sneha Gullapalli | | A100, AI infrastructure, AI Workloads, cloud native AI, Dynamic Resource Allocation, GPU autoscaling, GPU cost reduction, GPU optimization, GPU partitioning, GPU sharing, GPU time-slicing, GPU utilization, H100, Karpenter, KServe, Kubernetes DRA, Kubernetes GPU scheduling, LLM Inference, model caching, multi-instance GPU, NVIDIA GPU Operator, NVIDIA MIG, NVIDIA MPS, scale-to-zero, VRAM
Stop Treating GPUs Like Web Pods
Kubernetes schedules accelerators as opaque integers, and your bill pays for it. Share the silicon, scale on the right signal and keep weights out of the image ...
Veera Ravindra Divi | | AI infrastructure, AI serving, autoscaling, cloud costs, cloud native AI, DCGM exporter, DRA, Dynamic Resource Allocation, GPU costs, GPU scheduling, GPU sharing, GPU utilization, GPUs, inference workloads, KEDA, kubernetes, Kubernetes GPU scheduling, LLM Inference, MIG, model weights, MPS, NVIDIA GPUs, NVIDIA MIG, Prometheus, scale-to-zero, time-slicing
A Green Kubernetes Deployment Does Not Mean a Healthy Application
The deployment finishes, kubectl rollout status reports success, and every pod shows Running and Ready. For most teams, that is the moment the release is considered done. Then a customer transaction fails ...
The Foundation Was Already Poured
Techstrong's Experts Exchange this October, Cloud Native Now: The AI Stack, and this November's KubeCon in Salt Lake City are both making the same case for cloud native and AI. The argument ...
Alan Shimel | | agent governance, agent identity, agentic AI, AI agents, AI governance, AI infrastructure, AI security, AI stack, AI strategy, AI Workloads, cloud native, cloud native developers, cncf, enterprise AI, GPU scheduling, KubeCon, kubernetes, Kubernetes AI, MLOps, model serving, observability, platform engineering, sigstore, SLSA, software supply chain security
Rust Rewrite Readies Kata Containers for Agent Sandboxing
The OpenInfra Foundation has released version 4 of Kata Containers, which arrives with a new Rust-based default runtime that brings new memory safety and performance gains. With the Rust rewrite, the Foundation ...
Inside the Packet: How Kubernetes Networking Actually Works at L3/L4
Stop guessing why your Pods can't communicate. This deep dive strips away the magic of Kubernetes networking, providing a layer-by-layer breakdown of the dataplane using real packet captures, kernel data structures and ...
Supermicro Debuts Kubernetes Edge AI Appliance with Red Hat and Everpure
Supermicro has introduced a prevalidated Kubernetes Edge AI appliance developed in partnership with Red Hat and Everpure, aiming to streamline the deployment of AI applications across distributed edge environments. The new release ...
What Preparing for CKS Taught Me About Kubernetes Security
Preparing for CKS reinforced that Kubernetes security is not about a single tool or feature. It is about good defaults, clear boundaries and consistent operational practices ...
Security Flaw in Argo CD Can Let Attackers Take Over Kubernetes Clusters
Argo CD has become a widely popular open source tool for developers who use GitOps for deploying cloud-native applications to Kubernetes. For cyberthreat actors who are already increasingly focusing their attention on ...
Linkerd 2.20, the Latest Release of the Cloud-Native Service Mesh, Arrives
Linkerd 2.20 maintainer Buoyant claims it has slashed control plane memory usage by 85% in this version. If you want a lean, mean cloud-native mesh, Linkerd is for you. This latest release ...

