Hetzner continues its growth in the US with a new location
hetzner.com
hetzner.com
The UX behind managing instances is delightfully pleasant where new instances are available faster than any other cloud provider we've used, within seconds of creating an instance you can immediately login with your configured SSH keys. Another nice feature is being able to "rescale" your instance to higher specs after a restart [2], so you can confidentially start with a small instance that just meets your current workload knowing that you can easily scale up your instances as your workload increases.
AWS RDS was the only critical service keeping us on AWS, a service we no longer need in our new Apps which we're building with SQLite thanks to the effortless replication in Litestream [3] that we're using to replicate to Cloudflare R2 - another great value S3 alternative with $0 egress fees [4] where you can get even greater value & performance when hosting behind their free CDN.
[1] https://servicestack.net/blog/finding-best-us-value-cloud-pr...
[2] https://bizanosa.com/how-to-upgrade-resize-hetzner-cloud-ser...
Edit: And if you do manged Kubernetes and managed Kafka Instances the number would go up to like 95%. Oh and those Videos with der8auer? Really awesome to see, do Linus Tech Tipps or Level1Techs next!
https://github.com/kubermatic/kubeone/tree/main/examples/ter...
Though I'm sure if you ask and they have the money for it lying around, Kubermatic will sell you an SLA.
For folks wanting to learn more, we have this nice getting started tutorial covering all the cloud providers we support including Hetzner: https://docs.kubermatic.com/kubeone/v1.5/tutorials/creating-...
And for folks asking about SLAs, support, and stuff like that, I recommend checking out our KubeOne Managed Offering: https://www.kubermatic.com/products/managed-kubermatic-kuber...
I'm also happy to answer any questions y'all might have. :)
Proof: https://forum.hetzner.com/index.php?thread/21421-docker-cont...
You can find the link in my profile or here[0]
[0]: https://nimbusws.com
I have seen the "apps" in image selection in the Hetzner cloud console, so I'm convinced they're at least thinking about going up the value chain.
Would you have wanted a hypothetical discussion though (directed at grandparent), I would be up for that. Email in my profile
If we're taking requests, I would love a Fargate-like service. It would be absolutely fantastic if Hetzner provided a service where users uploaded a Docker image, and Hetzner took care of handling the networking bits and running it, and gather metrics.
Upload a docker image, specify container size (1cpu 2gb)
go live
scale from 1rps to 1000 rps any time
stateless
pay per request or pay per container
AppRunner, Fargate and Cloud Run have different ergonomics and specifics but thanks for this outline. v1 is likely to be pay per container but other than that feels doable.
Superb!
> would you mind telling me a bit about the metrics and the scale you'd expect?
It will depend mostly on what the service offers.
If the service only supports running a single isolated container without any scaling whatsoever then it would be helpful if we could monitor básico stuff like CPU and memory utilization, and also network traffic, free disk space, and also disk IO. If the service supports auto-scaling then it would be helpful to track all resource utilization rates along with all alarms and events involved. Auto-scaling also implies load balancing thus if that's the case then it would also be helpful to track the basic load balancing indicators, as well as request logs.
In the end it really depends on what services you're planning on offering, and how you'll charge for it. As a user I would need to monitor any metric which is directly and indirectly involved in determining cost, and on top of that I need to monitor performance.
> Would you expect always on or more of an ephemeral container?
The most pressing need would be always on containers to be able to go the lift-and-shift onboarding route to managed services, but ephemeral containers sound like function-as-a-service and those are pretty exciting as well.
Free disk space would imply elastic block storage or something similar so I’ll need to give that a think!
V1 is very likely to be always on so great that it’s the core use case for you!
Disclaimer: I'm the CTO
I don't understand their price page. They claim $0 egress fees, but their free "Class A operations (mutate state)" and "Class B operations (read state)" have a mothly cap. After that you pay by the number. Isn't that an egress free?
One time I hosted a new movie (mkv, 23gb+), that cost like $17 to “rent” on Amazon prime… so my extended family could just take the link, download it, or just watch it in their mobile browser or laptop.
The egress of the streaming alone (not downloading), while they aircasted, definitely adds up much more than one would expect!
S3 is charging $0.023 per GB, right?
IIUC - there's not a fixed cost for sending 1 GB of data. The cost differs based on where you send from and send to. So it'd be hard to have a really good estimate - but I'm wondering if anyone has a good ballpark figure.
My understanding is that transferring data on the same continent (vast majority of traffic) should be <$0.0005 - meaning the margin is really high.
I know it's more in Amazon's interest for your cloud usage to be as inefficient as possible - so they can charge you as much as possible and get as much margin as possible.
However, this product doesn't seem to make sense to me.
Why would you even pick your S3 regions? Shouldn't AWS balance your data for you across continents so that your data egress is automatically optimized?
Is that how Cloudflare works?
Bandwith cost highly depends on the scale and target network. Cloudflare has a good blogpost on relative bandwith cost: https://blog.cloudflare.com/bandwidth-costs-around-the-world...
To put numbers to the relativity, for a somewhat smaller datacenter in Central Europe with a 100Gbit/s connection, the running Cost of transit is somewhere around 0.00004€ per GB, on a theoretically fully sustained connection(not realistic, also with no redundancy or hardware). Peering is basically free(after initial buildout) with around 3k€ per 100Gbit/s (https://www.ams-ix.net/ams/pricing). This all excludes cost of investment for initial buildout and hardware. On a scale on amazons level this becomes even cheaper, since operating your own network is cheaper than buying transit.
Hetzner charges 1,19€ per TB, amazon 90$(on the first 10 TB). So amazon probably doesnt really care about the price differences per region and went with a mixed calculation, since their margins are so absurdly huge.
also, major parts of the cost are in hardware and staffing/development, of which amazon has way more, due to the demands of SDN.
No thanks, I'll keep using Digital Ocean or someone who doesn't make me jump through hoops.
The only thing I imagine someone could do with the copy of my passport is pretend to be me when they sign up for a similar service.
Recently I signed up for something else from Hetzner (needed temporary storage) and they didn't request anything even though I had closed my account before so I created a new one. So perhaps they have mended their ways.
I emphasize that I personally did not run any such tests, yet. But was wondering, since this is Hetzner thread, that maybe someone can share their experience, in particular comparing AWS's gp2/gp3 based deployments vs Hetzner's volumes.
I would also add that on AWS, for example, you can nearly seamlessly expand EBS volumes in size without downtime. Last time I checked, not an option on Hetzner - you must take care of expanding the file system yourself. Which makes sense, as Hetzner is more basic service, but it's worth remembering that there are various differences like that, when comparing the day to day operations between such providers.
Network volumes always have lower iops than local disks. Thats expected, except when you pay a huge price.
> I emphasize that I personally did not run any such tests, yet. But was wondering, since this is Hetzner thread, that maybe someone can share their experience, in particular comparing AWS's gp2/gp3 based deployments vs Hetzner's volumes.
AWS GP3 has a baseline performance of 3000 iops, and from there on you have to pay for each iop/s. To reach the performance of hetzner volumes you have to add 20$/month. But with AWS you can get higher than the hetzner value. But the iops performance of the dedicated machine is not possible with gp3 (max. 16k, compared to hetzner 30-40k for local disks).
With io2 devices you can get higher than 16k iops, and higher than the hetzner local disk.
Hetzner Cloud pricing is great, but their dedicated servers are even cheaper. For example, they offer a 16-core (32-vCPU) Ryzen 9 5950X for €103 with 128GB RAM and 2 x 3.84 TB NVMe SSDs. They offer a cloud server with 32 vCPU, 128GB RAM, and 600GB of storage for €296 - nearly triple the price and those CPU cores are probably not as good since they're likely Zen2 cores rather than the Zen3 cores of the Ryzen 9 5950X.
https://www.cpubenchmark.net/cpu.php?cpu=AMD+EPYC+7502P
https://www.cpubenchmark.net/cpu.php?cpu=AMD+Ryzen+9+5950X
I know, benchmarks aren't everything, but the Ryzen is getting 1,432 per vCPU while the EPYC is getting 766.
The AX51 costs €59 for 8 Zen2 cores (16 vCPU) and 64GB RAM while the CCX41 costs €154 for 16 vCPU and 64GB RAM.
I know, the cloud servers come with flexibility, hourly billing, and no set-up fees. I also know that their cloud pricing is very good. Still, I wish I could get a few AX101s in the US. 3 AX101 servers would be €310/mo, each with 16 Zen3 cores, 128GB RAM, and 2x 3.8TB of storage.
The most recent and egregious example would be banning users who run cryptocurrency-related nodes, even if not mining, despite only mining and similar activities being prohibited in the ToS. Here's a reddit comment from the official account stating that trading is also prohibited: https://old.reddit.com/r/hetzner/comments/wucxs4/is_it_allow...
While banning mining is perfectly understandable (the data centers are likely built around an assumption of normal usage patterns and not 100% CPU 100% of the time, and at least one cryptocurrency is infamous for wearing out SSDs with its mining), banning trading bots, proof of stake, and the operation of nodes for blockchain analysis is extremely surprising, even if you have read the TOS. And like in the earlier cases I've heard about, the servers seem to have been blocked without warning.
But preventing me from running a normal bitcoin node on my server, which writes 400GB once and is done makes me left confused.
If these are not toy projects, outages can get expensive (due to networks imposing penalties on nodes that don't fulfill promises, or trading getting messed up).
What will they do when a new form of fraud or a more potent way to do cyberattacks targets another online activity? Like ecommerce? Or blogs? Shut down all ecommerce sites? All blogs?
I find such measures as foolish as blocking port 25 - it works until the attackers and the software they use update itself to abuse the same thing in another way. A temporary relief that just complicates things.
Shows why you somehow think running a node makes it a target for cyber attack and frauds...
How does running a node make it a target for such? They only relay publicly available information and targeting a single node doesn't really harm the network.
FWIW, I've been running a seedbox in Hetzner's Falkenstein datacenter for yrears without issue. I only use private trackers + 1 public tracker focused on asian content, but they're definitely capable of figuring out that the my traffic is mostly torrent-related and have never taken action. The only time they've ever null-routed me was when I accidentally left a DNS resolver open to the world.
Seriously love Hetzner, amazing prices and really fast human support.
I'd say if you want to run gray area services like crypto, porn or gambling, you should probably ensure in advance that the provider is okay with you running your service, not just that they don't explicitly list your use case in their ToS.
The problem with building a platform that distinguishes itself by not having specific rules is that it attracts primarily people looking to do things that would break those rules and that this creates an environment that can drive away anyone else. This is as true for business services as for anything else.
The cryptocurrency market is currently already collapsing under its own weight. I don't think the idea should be attacked directly though (ideas are bulletproof), rather we should assess the side effects and ensure any harm is repaired. So if there was an opportunity to take direct action, it should have been at a regulatory level: carbon tax on the electricity, treat and regulate exchanges the same way you'd treat "brick&mortar" banks/exchanges/transfer services, tax mined coins as income, etc. GPU OEMs should've limited availability per buyer (no normal person needs more than 2 high-end gaming GPUs if they're actually just gaming). Even if you're a cloud provider, you should just charge extra for any excess wear on the hardware, maybe co-locate the "hot" nodes together, and leave politics at the door.
So on one hand, this is a form of censorship. I can imagine that given a sufficiently broad definition of a blockchain, you could use it to shut down any distributed, log-structured database project. On the other hand, the service provider also has the full right to refuse service to anyone, no explanation necessary - I'm certainly happy my company did so, whenever approached by any cryptobros.
It may soon be legal to not allow certain races to be your client.
I am not being antagonistic. I am quite serious.
I myself have built out my own private on premise data center and have no contractual restrictions on what I can do. I’m only limited by US law.
It was expensive . I did it because I didn’t want to be restricted by Ovh etc lease terms .
It’s the same as owning your house vs renting.
I came here to say the same thing. I've had production servers shut down by Hetzner with zero warning. And I believe they also kept the hardware running and continued charging you for it...just without network. As a result of this experience, I now consider Hetzner unusable for production systems.
Yep, that's the fault of Hetzner that prevents them from being used by many startups instead of AWS, Azure GCP etc. AWS is very tolerant of what is being done on its network, for example. Whereas a startup that uses Hetzner services will have to make sure that their use case is not prohibited by Hetzner. Moreover, they would have to be vigilant to keep following the changes to Hetzner ToS and practices to avoid getting disconnected out of the blue one morning. With restrictive German laws and increasing Eu regulations, that is a major liability for any startup.
It would be great if Hetzner had a full fledged US subsidiary that was subject to US laws - one that didnt have to get burdened by German law.
And those failed communications between provider and customer are why everyone is frustrated with the big corpos (google, cloudflare, stripe, etc.) which Hetzner is apparently bound to become.
This is not to excuse bad support experiences. I just think we should appreciate it when there is still a channel to a real human being willing to be helpful and engage constructively, because it's getting increasingly rare these days.
Hetzner is on my radar for a long time as an alternative to simpler deployments on AWS. It's great they are monitoring internet forums and try to be helpful, but also very frustrating to read such problems apparently still occur.
We recently had a mission- and time-critical issue with play store and our first reaction was to contact our gcp account manager and old google friends because we knew play store communication sucks. It should not be this way.
Let's hope I don't need another "one-time" server replacement because of your faulty servers, because I already used mine. 10 hours of downtime with angry customers on my tail sounds fun.
Doesn't surprise me, German companies always know better and will try to prove you wrong, whether you're their customer or supplier. Such is the business culture, I think.
Kudos for assisting your colleagues with better communication in English, I think this practice could be useful for many organizations with people that speak English as a foreign language.
My impression is that in Germany you bear the burden of proof that you actually have a problem, that its not your own fault, and that you did your part trying to fix it.
You usually can get good support in Germany ( even at government offices ) , IF you show up with your Leitordner ( legendary German ring binders) with all the receipts, all the possibly relevant account numbers, transaction ids and a detailed analysis of your own problem.
There is nothing that signals to German support staff that you have to be taken seriously like a ring binder, preferably with color coded markers at the margins and lots of punch pockets.
If you think this is satire, try it the next time you have an in person appointment...
The rest is entirely correct, though.
Yep. Its a plight that afflicts almost all engineers around the world, and German companies are even more afflicted by it due to the German engineering culture...
I chalked it up to our decision to run our databases on consumer grade hardware and didn't give it a second thought.
Yep. The attitude of the engineering support is something that Hetzner needs to improve a LOT on. They tend to treat customers as if they are treating members of their own open source or engineering community - scolding the customers when they think the customer (community member) is being unreasonable etc.
Its a great thing that eng. support is just a ticket and 20-45 minutes away at Hetzner. But the attitude needs a lot of improvement. Imagine that you are a startup facing some quirky issue that affects your business in the middle of the night and having to deal with attitude and scolds from datacenter engineers...
Since the EU datacenters seem to be operated by EU companies and the US company is merely a sibling subsidiary of Ensoxx, which itself is also an EU company, this should provide sufficient isolation to prevent interference from US agencies short of direct sabotage or espionage (since the EU staff is not in the chain of command of the US company).
So for a definitive answer you probably want your lawyers to talk to Hetzner's lawyers but at face value this is at least miles ahead of any US-based cloud provider, which in all honesty is still the default solution for most EU-based companies despite this ruling.
Conclusion:
In summary, you as a customer do have influence - to a certain extent - on shaping who has access to the data on your servers. EU and US authorities do have to follow the laws and legal procedures in requesting data. However, this may give you a false sense of security since some authorities have been known to stretch or violate agreements. If you require a web hosting company that has absolutely no connections to the USA, then unfortunately, we may no longer be the best choice for you. Since Hetzner US LLC is part of the Hetzner Group, there certainly is a connection. We hope that we have explained things clearly from our point of view using the two above case studies.
So, because Hetzner is not owned by a US company, stuff like the CLOUD act doesn't apply to them. So, if you have a contract with the German entity of Hetzner and use a German server, you should be fine in terms of GDPR.
Any of the big cloud providers can claim that they comply with EU legislation, but they also have to comply with US-legislation and if 3-letter agency wants to have some data from one of their subsidiaries in EU, then they can/will decide which contract to breach.
I read Hetzners statements as being that they can no longer guarantee that they will not be forced to do the same - but that can be my reading of their statement that is wrong.
If I already had them as hosting-partner for a solution that fell under Schrems II, I would have them confirm this, to be sure.
Hetzner US does not have a European subsidary and therefore cannot violate GDPR (assuming US personal can't access EU customer data).
Hetzner HQ is in Germany and is not allowed to enforce the CLOUD Act outside the US
But if I was under legal/contractual obligations, with Hetzner as my hosting provider, I would have their legal department confirm this.
Since Hetzner found the need for appending the paragraph I referenced, they must have become aware of something.
Now that they are entangled with US law there might be an incentive to be as a cooperative as possible.
Yet, Hetzner is still a "better" option (with regards to data protection) than any of the big US-based cloud providers.
Imho, as soon as you do business with the US or trade in US Dollars, you need to play nice with the relevant authorities.
If I understood it correctly, Hetzner is now "infected" in the same way as the three US cloud providers are. The Schrems II verdict and Cloud ACT basically concludes that no European company can exist in the US and vice versa without having to deal with the same pesky legislation.
An alternative could of course be that Hetzner created a new US based company where the EU parent Hetzner company only holds a minority ownership in the new US-based company. The EU based parent company in turn then "sells" its technology to the new US company. This way, the arrangement becomes more reminiscent of how IBM has sold its mainframe to European companies...
What makes you think that a European company operating within US jurisdiction would not be subject to the same laws?
If the European company receives a request from the US authorities for information, they need to follow the same legislation as the US companies do. Just because it's a subsidiary won't help. The authority will say "we want to know everything you know about the following person, please give us the information, otherwise...". The authority will not distinguish whether it is a subsidiary or the parent company.
Of course have the choice to just ignore the request from US authorities, but then you have to be aware of the consequences, i.e. quickly give up and shut down the subsidiary and stop trading with US dollars.
This is the root of the problem. CLOUD act has been ruled illegal in the EU just as you said, but it is also illegal not to comply with CLOUD act in the US. And companies operating on both continents in practice need to comply with both laws, regardless of whether it is a parent company or a subsidiary.
At least that's how I interpret it...
But it really depends on how infectious just owning a company is, which I have no idea. But my gut-feeling is that it shouldn't be too infectious, since otherwise just buying a single share of a company operating in another country would put you into legal peril (who controls the subsidiary here is not really relevant, since the Cloud act wants to swim in the opposite direction in your scenario, therefore it shouldn't matter if it's 0.01% or 100%).
Hetzner Europe is owned by Hetzner Group, a German company. Hetzner US is also owned by that German company. Hetzner Europe isn't owned by a US company, it's just a sibling to one.
In the 2022 annual financial statement they do mention expanding to the US, though they don't go into the legal details. As the link posted in the cousin comment mentions, though:
> Hetzner US LLC, as a subsidiary of Hetzner Online GmbH, provides data center services within the USA for the parent company, Hetzner Online.[2]
So there is no US owner.
[0]: https://www.northdata.de/Hetzner+Online+GmbH,+Gunzenhausen/A...
[1]: https://www.unternehmensregister.de/ (enter "ENSoXX Holding AG" in the text field)
[2]: https://docs.hetzner.com/general/general-terms-and-condition...
78M profit on 290M revenue
(See [0] above)
_ edited figures _
There are some Youtube videos of people getting tours of their data centers. It's a very custom setup, keeping both component costs and energy costs down. Their earliest servers were basically tower PCs on shelves, their more recent generations are more akin to custom rack designs with inhouse-assembled servers, with a datacenter design that exploits natural convection to do a lot of the cooling.
a friend of mine told some support staff got 20k EUR in bonuses last year because of that
probably they figured out, it's easier to give some of the profits away than pay a higher tax
Choices for EU only companies are getting smaller, and this is a real let-down.
Also, track record of at least some of them is not the best, if we are to believe, for example, comments on HN. I've made initial research about most of them. Just to give some examples about Scaleaway: prices moving up by 75% without notice (https://news.ycombinator.com/item?id=25999148), zero notice of the removal of the ability to start ARM64 instances (https://news.ycombinator.com/item?id=22923413).
If taking requests... showing memory charts along with the usual CPU, network bandwidth plots woudl be very helpful. :)
As it was going to host (among other things) my mail server, I could not use my primary (self-hosted) email address - because that's asking for trouble if I ever run into issues. It seems using Protonmail triggered something on their side.
They asked me for a copy of my ID card, which I happily provided. But they still refused my account, without explanation. Oh well, their loss. I've been at OVH now for a few years without any issues. OVH's product is definitely worse, but at least they'll actually let you use them.
Clearly if you are rejecting legitimate customers for arbitrary reasons there’s still some work to do on your approval process.
I mean, I get that there are bad actors, and that it’s rationally better for Hetzner to have false positives than false negatives. But it just feels wrong. Legitimate customers shouldn’t be rejected until they prove they’re malicious.
Low-cost attracts more fraudulent customers, using stolen credit card numbers. Hosting is particularly bad for this compared with other low-cost services, because of the community of people who want to use rented servers for DDOS and such, ideally without paying or being tied to a real identity.
It also attracts people who will do a card chargeback if the server isn't what they wanted or after they've used it for some temporary event. Some people don't appreciate that chargebacks are expensive for the low-cost supplier, and some people don't care.
Low-cost also means the penalty cost of credit card chargebacks is a higher proportion of income, even if the number of them was the same. It might be so much higher that the business couldn't be profitable at the prices it offers if it didn't aggressively filter which customers it takes on.
Or would that just not scale?
That's their courteous treatment. They asked me to consent to an AI scanning my face!
edit: Specifically with this startup,
I too provided US State issued identification, which still resulted in my account being denied.
I generally think highly of the company, and want to pay them for their services; but they made it impossible for me.
The TOS are vague. You're not allowed to "violate the rights of third parties", but they don't have a clear policy on what is a violation. If you review a product and the company files a DMCA, they may or may not say that naming a brand is a violation (I've experienced them coming down on both sides of the issue), when it's clearly fair use.
For DMCA claims, they tell you to use a form to reply, but the responses from the form will not be received and/or read reliably, so half the time you get a follow up email after 24h saying you haven't replied, so they will now manually check resolution and might block your server.
Tech and pricing are fine and I recommend them, legal is an issue, and I'd include some "but only if you're in an industry where DMCA claims are unheard of" caveat with my recommendation.
This is like mentioning in your rule-book that metal toothpicks are forbidden because I don't know, people can short your batteries or whatever. But also banning wooden toothpicks without mentioning this in your rule-book.
Their ToS are not always vague but the way their terms are executed is definitely arbitrary.
Only thing that's missing are BGP sessions, would make it easier to build anycast networks with them. You can get it if you buy rackspace but I really don't want to manage my own servers anymore.
Another thing that's missing is geo-redundant load balancing, i.e. the ability to spin up a single logical load balancer in multiple regions and dynamically route client traffic to the nearest DC via BGP. Currently LB targets need to be in the same DC, which is kind of weird. You can use DNS load balancing but given the larger footprint they should really invest in a proper anycast setup, I think.
I'd definitely vote on that!
but you are hosting cloud offering on some physical servers already, so you can start selling these servers as dedicated..
I don't plan to switch but Hetzner offers a CPU optimized server with 4 CPU cores and 8 GB of memory for $17 USD / month. DigitalOcean offers the same thing for $56 USD / month. It's hard to ignore how much of a difference that is.
Even the smaller'ish instances have a huge difference. Hetzner has a 2 CPU core 4 GB of memory server for $6.75 USD / month. DO offers the same thing for $24 USD / month and that's using the worse grade CPU. If you pick the higher end AMD CPU it's $28 a month (which sounds similar to what Hetzen is using but ~4x less price). For perspective DO's $7 USD / month price point gives you 1 CPU and 1 GB of memory.
Edit: As someone brought up in the comments below, it's possible that the Hetzner price ends up being ~25% cheaper due to not charging VAT. This will depend where you live. I converted Euros to USD on Google based on their public pricing page https://www.hetzner.com/cloud as a US site visitor.
In one of my Java toy projects I got http responses to sub 10ms, that's includes, querying a small amount of data from postgres and responding as JSON. The tls handshakes might have been reused through, now that I think about it. The number was taken from the network tab while switching around routes in the pwa
The bare metal instances generally have nvme storage, so you get incredible IO
As a reference from a recent test: I went with the smallest OVH bare metal server that was discounted to 30€/month the other day (normal price 60€) and got about 2k with hdparm. Running the same test on my 45€ hetzner instance got me over 3.1k. And the hetzner instance also got twice the memory (64gb), though that didn't impact this particular metric.
[1] https://servicestack.net/blog/finding-best-us-value-cloud-pr...
That's good to know the prices are even more of a difference. Hopefully this sparks DO into being more competitively priced. If someone wants to throw up a few servers somewhere on the cloud and doesn't care about managed features, it's really hard not to consider using Hetzner for that.
I put in a ticket and within about half an hour they had installed a replacement disk. I then had to reinstall everything because I'd messed up my raid config but that was totally my fault not theirs.
I was very pleased with that efficiency on what is a very cheap server.
Wish they would add managed databases to their cloud.
I currently have hybrid DigitalOcean / Hetzner setup to take advantge of load balancer and managed database at DO. If Hetzner provided some of those, I'd gladly switch.
I am not managing replicated postgresql myself ever again.
Am I you?
I've been there for 2y, pretty happy with them so far.
+1 for managed Cloud DBs. I'm surprised they don't exist already, given that they have managed DBs on their cheap web hosting platform. Shouldn't be a big step to make that available in Cloud.
Another request I have is virtual routers, so I only need 1 dedicated IP address and can NAT (or whatever) everything else. I get that I can do this with a small Cloud Instance and a private network, but those things are a pain to manage and I'm sure Hetzner could do a better job than myself :-)
IT is basically 2 things, at its core:
* compute
* storage
Hetzner offers cheap compute, which is great because modern applications can have stateless web/app servers.
Now, the missing part is storage, and that's much harder and riskier since it's state, inherently. A server dies, you lose stuff if you misconfigure it.
So people want a fully managed IT solution.
It's simple and it's obvious conceptually, just hard.
And I guess Hetzner won't do it because it's hard thus expensive to do.
Postgres (and later some exciting flavors like Neon) is actually the only database I plan to support, at least for right now -- MySQL and MongoDB are not even on the roadmap really.
I might make an exception for RethinkDB since I loved it so dearly.
Don't get my hopes up... I haven't used RethinkDB in ages and have fully moved to Postgres though
But also...
Don't know who is behind it -- a company named MostlyTyped (not sure if this has old RDB engineers in it), but warms my heart that the repo is still seeing commits and usage.
I want hetzner to concentrate on the core capabilities: servers, storage.
I want Hetzner to provide a PLATFORM to many different database-as-a-service on top of Hetzner. I want choice, and I don't want Hetzner doing what AWS does: stifling competition by releaseing their own flavor-of-open-source-database that puts the core projects sponsoring the open source code out of business.
Hetzner has been my absolute favorite infrastructure/cloud provider to date, it's criminally underrated. They've also got a great list of community tutorials for new sysadmins to peruse[0].
I like Hetzner so much I think it deserves a more varied managed service offering, so I'm building it[0].
[0]: https://community.hetzner.com/tutorials
[1]: https://nimbusws.com
Very happy with both the cloud and dedicated servers, but it's weird how bad the communication is on the object storage thing. If it's not going to happen, why not say so?
Wasabi is a great, inexpensive object store for anyone who's looking for that. I've had a really good experience with them (we have a little over 80TB of data there), and have found their support to be top-notch especially when you consider the price.
It wouldn’t even needed to be integrated to their API and billing.
Storj pricing is basically unbeatable. $4/TB/month, $7/TB/month for egress bandwidth which (I heard, yet to try) can be saved further if you put something like cloudflare in front of your bucket.
Speaking as someone who set up their own minio cluster (on Hetzner) as a way to have object storage at the lowest cost possible, if Storj was an option 3 years ago, I would have saved me quite a bit of money and time.
[0]: https://storj.io
There are just two things I’m missing:
- snapshots for volumes!!!
- restricting the dns api token (I don’t want to give one acme solver full access to 20+ domains)
- Separate account per domain .. which is a lot of work, see acceptation process in other comments
- Use a NS record for _acme-challenge.domain.tld when having the DNS hosted elsewhere and point this to the Hetzner DNS serversThe link to the page with the new prices didn’t have some overview of the changes, and figuring out how much more I was going to pay was frustratingly hard.
Hetzner’s article[1] just clearly explained why they increased their prices and had a table with the new prices.
I really appreciate that no-bs way of communicating.
[0] https://www.digitalocean.com/blog/new-4-dollar-droplet-updat...
[1] https://docs.hetzner.com/general/others/price-adjustment/
So I had customers who weren’t able to visit my website hosted on DO. Their provider refused to unblock the IP range, because there were a lot of attacks coming from there. DO doesn’t seem to care a lot about reputation.
I signed up as a European company, it was quite easy for Hetzner to check that I’m a legit customer.
And that’s probably their target group. Companies in Europe and the US. Hobbyists who just run one cheap server for fun are not a huge profit, but high risk to be fraudulent.
The latency to the EU somewhat stops us using them for production-level work, but this deployment to the US (especially on the West Coast, how lucky) will definitely shift our spend from DigitalOcean (which has been garbage lately both in terms of customer service and product offerings being much worse than advertised) to them.
Hetzner's control panel and automated systems are pretty bad, but that's a small thing when the price is right and customer service is much better than other offerings.
EDIT: I am a consumer of dedicated servers from them only. I don't really believe in cloud-based / AWS-type services for small businesses, and would like to get off most of our VPSes as well.
You do you.
I went back and forth on this with support for several months. They were not able to reproduce (it was a rare fault, happening on average only about once every few weeks), and blamed me for 'overclocking'. Eventually I shut off the account.
Hetzner is cheap, but you're on your own if you need any form of help.
Unfortunately I just kept reacting to it by rebooting, instead of writing to Hetzner support.
When I did eventually write to them, they scheduled a date for complete replacement of the main hardware transferring over the SSDs without any fuss, and the problems have never reoccurred, so that was great.
The reason I didn't write to them for a long time is I'd heard stories like yours, and I thought I'd have to gather lots of evidence to prove the server was unreliable. I did that and presented it, but I was still surprised at how instantly they just proposed a date for replacement without any discussion.
I have my eye on one of those new ARM many-core servers, but I'm not going to pay the setup fee and month rent in advance only to measure the performance and find out quickly that it's not what I'm looking for (because the AMD line is really good).
You offer a refund policy by writing to support, but I always felt a bit dirty at the idea of using it, as though it's abusing the process if I know before I buy that I'm probably going to return the server.
What I suggested is a feature that might win you some more customers. I'd find it useful so I imagine some other customers would as well.
As a data point, my purchase of an Ampere is contingent on gathering more knowledge about it, so that's at least one customer who's hesitating over a purchase. I would happily pay double rent during a short test period - I'm not looking for a freebie.
Unfortunately, I don't think general benchmarks and other customer experiences are likely to tell me what I'm looking for. I'm working on a new compiler focused on performance of various workloads, which is why the Amperes seem interesting, because of their unusual system architecture. They are different from other ARMs, so (for example) experience with an Apple M1 or Amazon Graviton does not translate.
I'll probably end up renting another cloud provider's Ampere VM for a few days when I'm ready; this is not a blocker for me. I just wanted to share that it's a feature I would have found useful if Hetzner offered, and the same has been true with other Hetzner servers in the past.
Simple UI, certainly robust infra (for my case at least) and the best, by far, prices in the market.
I love their pricing. I’d love it even more if I could announce my IPs on their dedicated offerings.
I want multi node, multi AZ hetzner k8s clusters ;)
I already run cluster in GER and US and what has happened in the last years has been amazing, but its not all the way there just yet!
Changes needed on Hetzner took less than a day, like buying a new server from auction.
Changes through corporate took somewhere around a week.
P.S. Sorry for digression
I suspect it does cost more but i doubt it’s a “huge leap” but it also defines what you define as huge leap.
Much less than a day. I found new auction servers took only a few minutes to be up and running after selecting them. Same with default config new servers (no extra SSDs etc).
Last time I checked through Vultr[0] from South America, Ashburn was significantly faster than Digital Ocean in SFO. Though, not sure if this helps you.
A kind of interesting statement, given that Intel has a large presence in Hillsboro (>15K employees IIRC). Not that this would matter to Hetzner, but still curious.
Perhaps it could be an option to let people back up their words with money up front? I.e. put $10k in the account, proves you can probably afford more than 10 servers.
What I'd like to have is something like Linode's start scripts feature which I've always found to be very nicely implemented.
If you really really need startup scripts, it'll be trivial to add it to your own image. Packer (and similar tooling) makes it trivial to create your own images as well.
Hetzner offers "managed bare metal", for extremely competitive prices. But, in my experience, exactly what I'd expects: "pay peanuts, get peanuts". I love, and chose them for some proof of concepts, early phase and hobby projects. But would not choose them for anything that requires serious stability and availability.
Not because they have flakey or even unpredictable service, but because the trade-off is that "bare metal" requires more work done by me, more responsibility for me, less options for quick failover and so on.
Just like for some situations, a server in your attic is the perfect fit, yet for others a managed cloud infra is the perfect fit, hetzner has some sweet spots.
Since then they priced their boxes competitively relative to DO. A little bit less. That's still holding.
Also note that at that time hosting in Europe wasn't a popular option, unlike what we see now.
Dedicated servers there were an option way before DigitalOcean, if I remember correctly (I recall at one point having some Linode and some Hetzner servers, DigitalOcean was not in the picture).
Early example: a weirdly memorable ad captaign around 2000 that for many years occupied the single most expensive computer related print ad slot in Germany (decisive spending) with a series of ads that seemed not quite "high production values", but also not deliberately grungy, a weird "definitely trying to be high gloss perfection, but somehow not quite there" (like in-house best effort or some local design house, certainly not the big-name agency you'd expect for ads on that slot). More recent example: their hardware seems to be a continuation from early-Google style "desktops on shelves" that's now a custom rack design (still noticeably lower density than typical 16") that's all about finding good price/reliability spots in cheap CotS parts, e.g. according to certain "begins the scene" blogger visits they sort for publicity, price-optimized custom versions of desktop mainboards (same PCB but not placing any parts they don't need). Chances are company with big investor backing would either go all standard rack parts (from a supplier like Dell or something like that) or go all in designing their own.
If there was a similarly priced alternative I'd be gone in a second.
Their idk.. the guy who makes the systems and orders hardware is stuck in the past. Arno or what his name is. I'd rather have 4×2TB than 2×10TB HDDs. I'd rather have low powered cores but more cores than those beast machines. Anything below highest tier is a consumer computer. And that highest tier is expensive, more expensive than comparable alternatives from Leaseweb or similar providers. Connectivity is only good for the country it's in and the immediate neighbors. That's for Germany. The Helsinki DC has an unreliable connection to Europe mainland. 20ms-60ms-80ms latency is not what I'd consider stable.
They are greedy.
"If there was a similarly priced alternative I'd be gone in a second." Cool, so they're cheap, decently-ish on support and there are no alternatives, yet you complain they are making some money from people that don't upgrade their old hardware.
Of course now that’ll also increase to €50, but yeah, war.
However, whenever I price compare, I always find OVH and Kimsufi far under-price them on small servers.
Am I missing out on a Hetzner product line or are they offering features I'm not getting on OVH?
This in turn makes everyone beholden to the EU regulations and German censorship regulations. Yeah no thanks!
I've been with DO for years, but after some problems and lackluster/non-existent service recently, I'm considering moving off to another provider.
They're easily my favourite hosting provider!
But their recent reputation around "crypto" [except we make up what that means on the fly] mega-bans is toxic to any business that wants to use Hetzner for reliable hosting of anything mission critical.
Here's how I understand their reputation at the moment. (Clarification from Hetzner would be very welcome!):
A combination of not clearly saying what they will ban, giving zero notice, banning all of a customer's servers at once, blocking access to backup data as well, and providing no recourse to recover (e.g. by stopping running something). So at best they should be used as a cheap backup or compute resource, but it's an uncomfortable risk to use them for something mission critical - a mail server or customer data processing server for example - as long as they keep to this "intentionally vague surprise mega-ban" policy.
They don't seem to ban a lot of customers, but the recent uncertainty affects more. It's not good that a number of customers, including me, simply can't tell if they are at risk of a ban, especially when running novel complex services. Even if they just banned one server and let you carry on operations with the others, that would be a big improvement. Or if they told you what to stop doing. Or if they provided time to get your data off or to correct an issue to their satisfaction.
It's one thing to have a policy preferring nobody runs a blockchain node, or a news or software distribution website which is blockchain-adjacent. Similar perhaps to those places that have a policy against IRC or gaming nodes. It's another thing to not say what the real policy is anywhere and associate it with abrupt mega-bans. The ToS only says mining.
Advice I've seen to someone banned that "you should have taken backups, tough" is mocking: After a ban you can't access your backups which they encourage to keep on their backup service, of course. It also makes sense technically to backup onto another Hetzner server, maybe at a different datacenter. Unfortunately the only safe thing to do is backup outside Hetzner entirely but they won't recommend that until it's too late.
As far as I can tell, there is no place at Hetzner which says what their mega-ban policy really is, and it looks capricious and unpredictable in practice. Different Hetzner staff say different things. The few public responses on this show that they appear to not care to understand the question, which adds to that sense that you don't know what activities are a risk and what aren't. Part of the problem is that peer-to-peer distributed systems in general are being ever more relevant, and look like "crypto" from the outside (and crypto-related techniques underly some technical methods of stabilising p2p networks).
Someone who only runs a website they think is safe will get banned one day under that policy, because it has some crypto news or something on it, or because some Wordpress module uses a p2p client to fetch some files, and they will be completely surprised.
It reminds me of Google and Stripe, where we hear a trickle of randomly banned customers whose lives or businesses are ruined through no fault of their own, with no recourse.
Except as far as I can tell, unlike with Google and Stripe, complaining about a Hetzner ban on Hacker News seems unlikely to have an effect. There is no Edwin for Hetzner. Or is there? Maybe that will change now they have a USA presence :-)