Finally documenting my hardware setup. Two identical nodes running AMD EPYC processors.
Each node specs:
Total cluster: 48 cores, 512GB RAM, 32TB NVMe storage. Enterprise power, homelab budget.
With 50Gbit networking, live migrations are incredibly fast. I measured just 3ms of downtime during VM migrations between nodes.
The secret is memory pre-copy combined with high bandwidth. The VM’s memory gets transferred while it’s still running, and only the final dirty pages cause a brief pause.
This enables true zero-downtime maintenance of my Proxmox hosts.
It’s always DNS. This week I spent hours debugging connectivity issues that all traced back to DNS misconfigurations.
Common pitfalls I encountered:
Pro tip: Always check DNS first, even when you’re sure it’s something else.
Sysprep can be frustrating. After many failed attempts, I finally have a reliable Windows Server 2022 template process.
The key steps:
My template deploys a fully updated, domain-ready VM in under 3 minutes.
Network segmentation is not just for enterprises. I implemented 8 VLANs to separate traffic and improve security.
My VLAN layout:
Inter-VLAN routing happens on the firewall with strict ACLs.
