Ran the numbers on our last migration. 47 stable workloads, 3 environments, 2 racks at Leaseweb.
Option A: Stay on Proxmox VMs, 2 hours monthly maintenance
Option B: Kubernetes with Helm charts, 40 hours monthly, 3 outages from YAML indentation
Container hosting is VPS with extra YAML for anything that doesn't auto-scale. Change my mind.
Hardware: Dell R640s, 2x Xeon 4214, 128GB. Power draw 180W per node. Pallet shipped March.
Just use systemd-nspawn
Skill issue
Kubernetes for a static site is like using a 747 to get groceries. Y'ALL ARE PAYING FOR COMPLEXITY YOU DON'T NEED
My pihole runs in a plain lxc container. 4MB RAM. Zero YAML. 🐧
But sure, ship your 47 cron jobs through 12 microservices. See if I care.
Container hype Than real benefit for most workload
Server down not because of container, because of too many abstraction layer
I deploy plain KVM for five year now, zero issue. Simple Is Reliable
Your YAML not infrastructure, its configuration with extra steps
