Newsletter
Subscribe our newsletter
Get new infrastructure guides, comparison reports, and migration notes in your inbox.
Tag
Infrastructure
38 articles tagged Infrastructure, newest first.
Assistant vs Agent: Where Your AI Security Model Changes
Five properties turn an AI assistant into an agent. How each one adds attack surface, where the OWASP lists apply, and what controls work.
Excessive Agency Is a Permissions Problem Wearing an AI Costume
OWASP LLM03 Excessive Agency has three causes: too much functionality, permission and autonomy. Only one is the model's fault, and here is how to fix each.
Securing AI and Defending Against It Are Two Different Problems
"AI security" names two unrelated problems with different owners, different controls and different budgets. Teams that conflate them buy the wrong thing.
AI Cybersecurity: What 100+ Companies Want Now
OpenAI and 100+ signatories say AI attacks will spread fast and call for stronger access control, patching, testing, and defender tools now.
Proxmox VE Version History and Upgrade Paths (Updated for 9.2)
A running reference for Proxmox VE major versions, their Debian bases, known rough edges, and the one-major-at-a-time upgrade path to 9.2.
Zabbix Pros and Cons in 2026: Strengths, Weaknesses and Fit
The pros and cons of Zabbix in 2026: what it does well, where it struggles, and which teams should or shouldn't choose it.
Building Standardized Service Catalogs for Infrastructure and AI
How to turn compute, bare metal, storage, network, model deployment and API requests into catalog items with owners, quotas, approvals and metering.
The Datadog Line Item Eating Modern Infrastructure Budgets
Why Datadog observability spend is easy to underestimate and hard to unwind, and how engineering, finance and cheaper alternatives shape the debate.
I Upgraded My Servers From a Bus Ride: Proxmox 7 to 9 Went Smoothly
A long-delayed Proxmox 7 to 9 upgrade, started from a bus, went far smoother than expected. Why Debian-based upgrades work and where they still break.
My Homelab Network Diagram, Then and Now
An old ExcaliDraw homelab diagram shows a bigger setup of DL380 servers and 10-gig gear, and how real life reshaped it into a Portainer cluster.
Inside Vertiv: Working at the Data Center Giant
What working at Vertiv is like, from people in the field: heavy travel and workload for UPS technicians, decent culture in some offices, and fast learning.
From Kopia Error to BlinkDisk: Simpler Backups for a Home Lab
A Kopia repository server path error on a two-device home lab led to a switch to BlinkDisk. What went wrong, and why the simpler tool fit better.
Is VMware Dying? The Future for VMware Administrators
VMware admins fear for their careers as prices rise and companies migrate. What the numbers say, and why virtualization and automation skills carry over.
Uninstall It Now: Huntarr, TrueNAS and a Supply Chain Scare
How the Huntarr app was pulled from the TrueNAS catalog within minutes, and what the scare says about trusting FOSS, Docker images and curated catalogs.
Sharing One GPU Between VMs and LXC Containers in Proxmox
Why a GPU passed through to a Proxmox VM cannot also serve LXC containers, and the setups that do work: VM passthrough, host-shared LXC, or vGPU.
DNS Filter Bypass: How Devices Dodge AdGuard Home
Why your local DNS filter gets bypassed by DoH, DoT, hardcoded resolvers and apps, and how NAT redirects, port blocks and a local 8.8.8.8 win back control.
How to Run a Java Monolith on Kubernetes Without NodePort Hacks
A practical production guide to running a Java monolith on Kubernetes without fragile NodePort duct tape.
What Actually Load Balances Production Traffic in Kubernetes
How Kubernetes Services, Ingress controllers, MetalLB and service meshes split the work of load balancing, and where TLS terminates in production.
Should You Use CPU Limits in Kubernetes Production?
A grounded take on when CPU limits help, when they hurt, and how to choose based on workload behavior.
2,000+ Service Accounts, No Owners: The Multi-Cloud IAM Problem
Why unmanaged machine identities across AWS, Azure, and GCP become a security and governance crisis at scale.
Real Kubernetes Production Failures From Control Plane to IPs
A field report on real Kubernetes production failures and the human factors that trigger them.
5 Best Zabbix Alternatives in 2026 (Compared)
Zabbix alternatives compared for 2026: CloudSino for hardware control, Prometheus for Kubernetes, Datadog for SaaS observability, PRTG for SMBs and more.
Lost ESXi Root Password: Reinstall and Keep Your VMs
Lost the root password on VMware ESXi 7 or 8? Why there is no supported reset, and how to reinstall while keeping your VMFS datastores and VMs.
New Proxmox Tool PveSphere Makes Big Promises, Meets Skepticism
PveSphere launched as a production-ready multi-cluster manager for Proxmox VE. The community answered with cautious optimism, jokes and hard questions.
When a Three-Node Proxmox Cluster Becomes a Small Data Center
A three-node Proxmox cluster with 4.5TB of RAM and hundreds of CPU cores drew attention once readers saw it was serious production infrastructure.
AI Didn't Kill DevOps, but It Made the Stakes Way Higher
AI speeds up DevOps work, including the mistakes. Where AI helps, where it adds risk in production, and why humans must stay in the loop.
Zero-Downtime Deployments Without Kubernetes
Zero-downtime deploys without Kubernetes: load balancers, blue-green, symlink releases, SIGTERM draining, Docker setups and safe database migrations.
How One Feature File Caused Cloudflare's Worst Outage Since 2019
How a ClickHouse permissions change doubled a Bot Management feature file and caused Cloudflare's biggest outage since 2019 on November 18, 2025.
VMware's AI Integration Is Here, but Do Sysadmins Actually Want It?
VMware launched Intelligent Assist, an AI chatbot in vDefend. Why sysadmins distrust AI in production and where VMware AI could still earn a place.
Cloud First, Regret Later: IT Pros on What Happens After Migration
IT pros explain why cloud migrations often cost more: lift and shift, hidden fees, OPEX accounting, when cloud pays off, and de-cloudification.
Tailscale Was Down Again, and Here's What the Internet Had to Say
A Tailscale control plane outage left users locked out of the admin console, raised status page doubts and sent some toward Headscale and backup VPNs.
AWS GovCloud vs Commercial Cloud After the us-east-1 Outage
Why AWS GovCloud stayed up when us-east-1 went down, why some federal users still had problems, and what the outage says about cloud dependencies.
Inside AWS's October Outage and What Went Wrong
How a DynamoDB DNS race condition disrupted AWS us-east-1 for over 14 hours, hitting EC2, NLB, Lambda and Connect, and what AWS is changing.
Multi-Region Failover: Why It Is Harder Than the Diagrams Suggest
After the AWS US-East-1 outage, engineers Googled multi-region failover. Cost, vendor dependencies and unrehearsed plans make it harder than diagrams show.
AWS us-east-1 Outage: Why Concentration Risk Still Matters
A DynamoDB DNS failure took down AWS US-EAST-1 and 82 services, from Slack to Ring. Why one region is still the cloud's biggest single point of failure.
When the Cloud Breaks: How One AWS Outage Took Down Half the Internet
The October 20 AWS us-east-1 outage took down Amazon, Slack, Fortnite, Duolingo and more, and showed how much of the internet depends on one region.
From Enterprise Bloat to OSS: A Kubernetes Cost-Cutting Story
A team saved $100,000 by swapping an overpriced enterprise API gateway for Kong OSS, then kept finding the same kind of bloat in other stacks.
Open Source Is Free Until It's Not: CNCF and the Cost of Free
The internet runs on open source tools maintained by volunteers who might burn out or walk away at any time. What happens when 'free' stops being free?