I moved a broker's virtualization from Hyper-V to Proxmox with zero downtime

10 hours ago

The author migrated a financial infrastructure's virtualization from a proprietary hypervisor to Proxmox without downtime.
Key reasons for migration: cost savings, independence from vendor lock-in, and making infrastructure legible to tools like Git and AI.
The setup involved two sites with multiple servers and about twenty machines, aiming to consolidate onto a fresh Proxmox cluster.
Pre-migration cleanup reduced scope by removing obsolete machines.
A rolling migration strategy ensured no service interruption, with careful capacity planning to consolidate VMs.
Three migration methods: conversion for Windows VMs, rebuild for Linux containers, and fresh setup for domain controllers.
Shared storage issues with LVM thin required switching to NFS for live migration to work.
Future goal includes distributed storage like Ceph for improved resilience.
For the second site, backup and restore streamlined migrating thirteen services in two days.
Various technical traps emerged, such as network bonds, two-factor authentication, and container permissions.
Post-migration updates to DNS, monitoring, and security rules were critical for completion.
An incident due to filled storage highlighted the need for headroom and automatic management.
Outcome: cost savings, infrastructure versioned in Git, and enhanced control for a small team.

Hasty Briefsbeta