Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I recently tried out Hetzner Cloud, after using their dedicated servers for years, and have come away positively surprised.

It works well, and the pricing is very appealing. But it's not even remotely in the same league as AWS/GC/Azure.

All you get is DNS, private networks, load balancers VMs and storage volumes. That's it.

Sure, you can self-host Kubernetes, and they even have a CSI driver for mounting volumes and an official Terraform provider. Which is great.

But you miss out on the incredibly rich service offerings you get from the others.

It's a great start though, and I hope they keep expanding their offerings.



Well, essentually Herzer is missing PAAS offerings that most modern devs can't live without, especially managed databases.


Is "modern devs" the modern "No true Scotsman"? Plenty of developers and organizations get by without managed databases. Although in "7-day sprints on loop forever" world of startups, any effort to require less knowledge to run the stack is probably welcome, hence it seems "most devs can't live without" for a lot of things, like managed databases, containerization and yadda yadda.


Having a managed database is the sweet spot for productivity/efficiency between maintaining a full time database administrator and having engineers learn how to do it themselves.

How can a serious company using a relational database get by without paying for database expertise in some way or another?


Yes, of course once the company reaches a certain scale you need specialized roles and/or software that you use, that makes sense.

But is "companies of scale" what "modern devs" refer to? What is a "serious" company?

There is a lots of parameters that goes into the choice of self-hosted DB / managed DB, not just if the devs are "modern", or the company is "serious".


> especially managed databases.

There are other companies, though, that have great cloud-managed DB offerings. Herzer (I'm not previously familiar with it so don't know if they already do this) would do better to just partner with one of these companies in a "marketplace" like arrangement.

I was extremely close to switching to Aiven-managed postgres until GCP finally supported point-in-time-recovery on postgres.


> It works well, and the pricing is very appealing. But it's not even remotely in the same league as AWS/GC/Azure.

IMHO Hetzner is far above AWS/GC/Azure in price and simplicity.

The only drawback I see in Hetzner is that they don't have a lot of data centers with a relevant geographic dispersion, but that's only relevant for a Unicorn-level scale, or applications that require TP99.99 latency in the lower-end of the double-digit scale uniformly throughout the globe.


You skipped over without mentioning the gist of the comment you replied too, which was the large amount of additional services that those other companies offer, but Hetzner does not.


One person's rich services are another's vendor lock-in.


If you're referring to AWS's "serverless" offering, they are mostly managed instances of FLOSS offerings with the exception of dynamodb, and arguably are far from justifying the colossal premium they charge for using them. I mean, does anyone claim with a straight face that without AWS it's practically impossible to get a database or a message broker up and running?


It’s a completely different thing to „get it up and running” and to fully operate and monitor 24/7 a highly available environment using full time SREs with strict contracted SLAs.


Your comment is rather disingenuois and naive. If you have any experience whatsoever with AWS you wouldbe fully aware that AWS quite vocally advertises that reliability is the responsibility of the customer,not theirs. Theygo even further by stating in their "well-architected framework" that it's on you to handle redundancy with multiple deployments across separate availability zones. So your baseless assertions fly in the face of reason once you actually lay attention to the operational side, because as it's easy to see and understand it makes absolutely no difference if you parrot keywords all day because in the end reliability is on you, and if it's up to you anyway then that is not an argument in favour of vendor lock-in.


Yeah, they're really barebones. Do everything yourself, which is nice tbh. Last time I used them though, they had some draconian restrictions on mail lists. A few complaints and your server gets shut down. I was just sending newsletters to subscribers.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: