Migrating from DigitalOcean to Hetzner

Original Article ↗ Hacker News Discussion ↗

Cost savings and competition

  • Many see the migration as evidence that US-centric clouds and DO are overpriced, especially for steady-state workloads.
  • Commenters report large savings moving from AWS/Heroku/DO/Linode/OVH to Hetzner or similar EU providers; some say cloud markups are “casino-like.”
  • Others argue you must compare like with like: the article switches from a managed VPS to a single bare-metal box, so the price gap partly reflects a lower level of service.

Dedicated vs VPS / cloud: reliability and HA

  • Key tradeoff: dedicated servers are cheaper and faster, but hardware/network failures cause full outages until support intervenes.
  • VPS/cloud offers live migration, snapshots, easier scaling and upgrades, and sometimes better hardware redundancy.
  • Several stress that a single dedicated box with many apps and DBs is a big SPOF; proper HA would use at least 2–3 servers, replication, and load balancers.
  • Others counter that many workloads (blogs, small SaaS, internal tools) tolerate occasional downtime; simplicity and backups can be a better trade than FAANG-style HA.

Backups, databases, and migration techniques

  • Strong emphasis on real backup strategies: RAID is not a backup; replication isn’t protection against human error or corruption.
  • Tools mentioned: xtrabackup and mydumper/myloader for MySQL; Postgres replication with various HA tools; rsync, snapshots, and scripted cutovers.
  • Some critique the article’s “zero downtime” claims as under-specified around write blocking and replica promotion.

Hetzner pros and cons

  • Pros: very low prices, strong performance (bare metal, NVMe), good for K8s or Proxmox-style private clouds, simple console/API, European jurisdiction.
  • Cons: reports of harsh abuse/billing enforcement (sudden shutdowns/bans), ID verification frictions, perceived overselling on some products, and worries about IP reputation / DDoS protection. Others say their long-term uptime has been excellent.

Cloud lock‑in and architecture choices

  • Several plan or have executed moves away from AWS/major clouds due to pricing complexity, egress fees, and proprietary services.
  • Opinion splits between “pay for managed services and focus on features” vs. “build IaC + container/K8s stack once, then stay portable and run it on cheap providers.”
  • Some see modern tooling (IaC, containers, and even LLM-based assistants) as dramatically lowering migration and re-architecture costs.