Hetzner Cloud Goes US (Ashburn/VA)
In this technical retrospective (2021), @evilc00kie shares practical architectural lessons learned from running, scaling, or migrating production software under lean constraints.
Core Principles for Independent Builders:
Defaulting to simpler architectures (such as monolithic frameworks, SQLite, or managed VPS) prevents early engineering bankruptcy.
Evaluating true compute and egress costs reveals that high cloud bills can often be slashed by 80%+ through judicious re-platforming.
Code you write at 2 AM as a solo engineer must be easy to debug six months later without complex microservice dependencies.
Community Debate Themes:
HN engineers debated the tradeoffs between cloud convenience and raw server economics, database indexing strategies, and resisting premature architectural complexity.
CATEGORY Frugal Engineering →
HISTORICAL ERA Pandemic Bubble (2020–2022)
COMMUNITY SCORE 163 points
TOTAL DISCUSSIONS 112 comments
I've been working on bringing the managed services (which is still somewhat lacking though Hetzner has LoadBalancers) now to places like Hetzner and OVH. Right now I'm planning on starting with Redis/Memcached and S3.
If this seems like something you'd be interested in I'd love to hear from you (email in bio). If you don't have the time to write an email but maybe take a most-questions optional survey:
https://docs.google.com/forms/d/e/1FAIpQLSfyhBThpC2mYubNPQaKRDJhxgVtTphjl4WSSX5vMjYOMrAWtg/viewform?usp=sf_link ↗
I plan on getting a small landing page up to collect people who want to be first to try it out tonight/tomorrow at the latest.
I'm most excited about Hetzner going US because I think they're going to really shake up the stagnant (high) prices for dedicated compute in the US.
Any info on what DC they use? So far they've run everything themselves which made them slower to expand but was much cheaper to run. If they now use other DC providers it would explain why they only offer the higher margin cloud servers there. But would be curious if anyone knows where the data center is physically located. With their other ones they're also quite open about it so I was surprised they didn't mention it here.
EDIT: They do mention somewhere on their side that it's a third party DC. So probably no option to offer dedicated servers there (at least not at the same price as in their other locations). Would still be curious which one it is.
More Cloud providers is a good thing. But I hope more providers start following the method of an implementation of OpenStack that a customer can control. OpenStack has an API and tools which makes it much easier for a customer to start using cloud services in a standard way. Might not be good for "vendor lock-in", but it does let me ramp up on technology quicker. I've used OpenStack in a couple cloud providers so far and I really enjoyed it. (I've also administrated an OpenStack bare metal cluster, so I don't envy whoever is managing it at the cloud provider....)
"The host systems for our Cloud instances all have a redundant 10 Gbits connection. This connection is shared by all instances on the host. We do not offer bandwidth guarantees for our Cloud servers, but you can expect about 300-500 Mbits."
This sounds pretty underwhelming and probably what's preventing me from migrating infra to Ashburn. Other than that, looks pretty great bang for buck.
I like that tweet:
Bravo Hetzner, no chance for servers with private data (for example European customers) on it anymore, Exoscale it will be then in the future.