Why ever not? If you sell someone something but only let them use less than 80% of it, that feels like fraud.
It’s total bait and switch, and probably false advertising/fraud.
Obviously that's not as good marketing. Products usually look better if you lie about them to make them look better. But lying in marketing shouldn't be legal. And this shouldn't be controversial.
Maybe because the limits are not fixed and completely depend on the spending of the customer? If a customer is among the top 1% of the biggest users and – at the same time – among the top 1% of the smallest spenders, that simply raises the costs of the service for everyone else. There are only three choices here: a) get rid of the customer; b) make them pay more, or c) raise the prices of unlimited plans for everyone.
Let's say the provider has a 100 Gpbs upstream, 1000 clients, and offers a cheap plan where you pay for 1 Gbps with no limit. The provider obviously can't provide a sustained 1 Gbps to every client 24/7.
Now, how do you honestly advertise that? Because you can give people 1 Gbps service so long not too many clients do that at the same time. And you can tolerate people downloading 24/7 so long they don't all do it at the same time.
How much you have to crack down will depend on how close to capacity you are, and what your clients are doing.
But a service with a cap also isn't unlimited.
Making a pricing structure which works for the company and is easy to understand for the customer isn't easy. Nobody said it was easy. But I don't understand how you go from that, to arguing in favor of lying to the customer.
But the fact that there is a limit at all, means that you can definitely display it somewhere.
Suppose you have two users, John and Jane, John is in the 1% of most valuable customers, Jane signed up last month. The data center has plenty of free capacity in June, but is closer to capacity in July.
Now suppose that each user has their own dashboard for resources.
John:
June - Your bandwidth limit is 200 TB. Please contact customer support if you need to increase it.
July - Your bandwidth limit is 160 TB. Please contact customer support if you need to increase it.
Jane:
June - Your bandwidth limit is 40 TB. Options to increase this limit will become available after X$ in total spend.
July - Your bandwidth limit is 30 TB. Options to increase this limit will become available after X$ in total spend.
As a matter of fact, Hetzner already has something like that under their "Limits" section: Limits
Limits are set per customer and are counted across all projects that you created. If you need more resources, just send us a request, and we are happy to adjust them for you.
Outgoing traffic to ports 25 and 465 are blocked by default on all Cloud Servers. Send a request for unblocking these ports. Please note that we generally do not answer questions regarding limit increase on the telephone.
You don't necessarily have to advertise the fact that your limits will be dependent on lots of variables and will change (though you should disclose it). What you shouldn't do is say that there are no limits, which is clearly lying. Even if the lie only becomes relevant to a very small part of the clientele who'll be affected by this, it's still lying.Of course, that's not the world that we live in and "unlimited" offerings will be a mainstay of marketing in the industry, for the better or worse. Probably better for the org and worse for the clients. Kind of why I also hate mobile carriers being dishonest with their data plans. It's like nobody even knows what "unlimited" means.
In your scenario, 10:1 contention ratio, "up-to" 1Gbps throughput, 1Gbps link-speed.
In the UK British Telecom originally offered 'consumer' ADSL at 50:1 and 'business' (more expensive) at 20:1 contention ratio specifically to deal with the backhaul limitations.
If they advertised "you get a min of 200kbps, that could boost up to 1Gbps" that would be acceptable, as long as you actually got that 200kbps. The problem is they claim it is 1Gbps, then look at you funny if you expect 70% of what they claim.
For instance, OVH has previously terminated accounts of paying customers simply because it didn't want to do any business with people from an occupied territory[1].
Just don't call the plan "unlimited". Have well-defined limits. They don't have to be hard limits, they can be a gradual speed reduction after some amount of data per month. Just don't lie. I don't understand why you think this is a difficult concept.
BTW, if it was just "we don't want to do business with you any more", that depends on what your contract says. Most consumer ISP contracts allow an ISP to drop a customer for any reason without explanation.
Words have meanings for a reason.
The meaning of the word "unlimited" is not, in fact, "limited, but we won't tell you what the limit is — you'll get to find it out the hard way".
Whatever the terms of the "fair usage" policy are, that's what should be in the description.
Otherwise, is plain old lying and misreading — which for some reason you are OK with.
I'm proxying traffic through mobile proxies and expect nothing less than unlimited.
What herzer do amounts to false advertising when compared to the mobile traffic examples.
For example, Australia:
https://www.gizmodo.com.au/2022/06/spintel-acma-unlimited-si...
I'm surprised my own residential grade connection hasn't been flagged, given I consume over 10TB a month.
[1] https://old.reddit.com/r/nbn/comments/hwz7l4/abb_cancels_ser...
I think that what a lot of people have problems with, is the fact that Hetzner doesn't actually tell you what is okay and what isn't. Would using 70% of the link's capacity be okay? What about 60%? Or maybe 50% is where they draw the line?
Here's another host that I use, Time4VPS, which have more clear terms: https://www.time4vps.com/?affid=5294 (remove affid for non-affiliate link if you don't want me to get a kickback in case anyone buys the services; I also host my blog with them)
Their servers include a guaranteed amount of bandwidth, either at 100 Mbps or 1 Gbps. And should you exceed that limit, they're also clear about what will happen:
> What happens if I run out of bandwidth?
> We reduce your VPS server’s port speed 10 times until the new month starts. No worries, we won’t charge any extra fees or suspend your services.
So if I'd exceed the amount that's given to me with any of their services, my 1 Gbps link would become a 100 Mbps link, or my 100 Mbps link would become a 10 Mbps link. On a technical level, it's obviously worse than Hetzner's offering (speaking of how much data you can actually move), but the transparency is nice and I wouldn't ever have to worry about my services being suspended because they don't automatically enforce any limitations just so they'd get to claim "unlimited" in their marketing materials.
Disclaimer: I still use Hetzner personally, they're one of the more affordable platforms (pretty much the same pricing tier as Time4VPS) out there. Though it's also nice when service providers have some provisions in place to automatically introduce limits for those things that could cause problems, like the amount of bandwidth that's available to you. Or just charge you for it outright, so there are fewer unknown variables.
Much like the Hetzner article, there's an obvious physical cap that can't be exceeded, and they're supposedly selling up to that cap.
Even in a private data center, you’d typically move a workload like that to avoid disruption to neighbors.
"We had a good team on paper. Unfortunately, the game was played on grass"
There are so many ways in our society where there's unwritten expectations. In this case, the unwritten expectation was that they weren't going to meter you and you didn't have to worry about overage charges, but that the 1Gbps should be thought of more as a burst rate than a 24/7 sustained rate. Likewise, I'd interpret it to mean that if I had a heavy month, I'd be fine.
If I'm a student paying for a dorm in college, should I be allowed to set up a mining rig pulling 10 kW and costing around $720/mo to run? Electricity is included in my dorm fee! I'm paying for it and I'm going to use it!
This is kinda why AWS nickle-and-dimes customers for every little thing - they had people who would find any area of weakness in their pricing model and go after it. I'm not even saying that people did it maliciously. Jungle Disk was a really cool idea: a mountable disk that used S3 to store the files. The problem was that it started doing loads of S3 operations. At the time, AWS didn't charge for operations and now there's this popular program doing a lot of operations compared to its storage and transfer charges.
As someone that has created SaaS software, some customer always comes along and finds an interesting use of your software. Sometimes it's a customer not really understanding the product. Other times a customer has found a weakness in the pricing model and they want to exploit it.
I get it the attitude. It's a war between you the buyer and the company selling a product to you. Personally, I'd rather have a more amicable relationship in my dealings. Yes, that means having realistic expectations about offerings including what might be marketing "puffery" (whether it meets the legal definition of puffery or not; https://en.wikipedia.org/wiki/Puffery). I'm not going to say that you're wrong to want 100% of the link 24/7, but then you're just talking about being in a situation where the company exercises their right to terminate your account (which you would have agreed to when signing up). Looking at Hetzner a bit more, it looks like they charge €1 per TB or $0.001/GB on their plans that have overages - 10x better than cheap places like Digital Ocean.
Yea, there's lots of places in this world where you can cause headaches for other people by using something in a way it wasn't intended that you're entitled to do - but you're going to have an unhappy time fighting for what you're due all the time. Instead, I'd rather focus on using things the way they were intended such that I don't create adversarial relationships with those I work with. Yes, in this case it would mean that I'd probably buy a few servers or just offer to pay them €200 for the extra bandwidth. Or I could fight, they could give me 30-day notice, and I could find another "unlimited" provider until they see me as a toxic customer and I start it all over again. Again, you're not wrong. You're just going to have an unhappy time trying to enforce your rights and get something better than the company expected to be giving you.
And of all the scummy things out there, Hetzner seems to be pretty decent. €0.001/GB overage on their limited plans is really cheap even compared to good deals like Linode and Digital Ocean (which charge 10x that). Their servers are really cheap. Their pricing is mostly very straightforward - maybe with this being an exception. I'll take a nebulous "it's unlimited, but don't push your luck" over what most hosting providers are offering. I could probably do 250TB for a month and they'd ignore it if it wasn't consistent. I can transfer way more than I get Linode or Digital Ocean and they'd ignore it. Maybe you just want me to say that you're right and that Hetzner is being evil and the person is due their 250TB of transfer every month, but I'd rather go through the world thinking that things are run by humans who want to do the right thing and sometimes don't anticipate all the consequences and then something like this happens - especially when the company doesn't seem to be doing evil stuff. Like, we've all seen surprise AWS bills in the tens of thousands without enormous usage. In this case, we're talking about a pretty small bill compared to the usage.
Yep, that person has the right to use that bandwidth - and Hetzner has the right to terminate a customer for any reason with 30 days notice. I'd just rather try and work with a company in a reasonable fashion if they're willing to work with me in a reasonable fashion. The alternative is becoming the customer that the company hates.
Actually the written expectation was "unlimited" - without limitation. These so called "unwritten limitations" exist solely in your head, and are not suitable for companies to assume users inherently know.
They do not state that anywhere in their advertising or terms of service.
They are advertising 1gbps/24/7/month as implied by their "Unlimited" terminology this is a perfectly measurable value and then cutting people off if they use 80% of that limit.
While I wholeheartedly understand that they can't offer that level of service for all customers and they expect most user to not use the full rate 24/7 and price their products as such.
If I'm being sold 1tb of transfer and am cut off that 800gb that is unacceptable, that is what's happening here (give or take some gigabytes).
If this is a problem for them then they need to change their advertising or terms orservice to state their Fair Use or AUP to list these limits.
In this case I suspect they assumed reasonable interpretation and customer interest: No realistic hosting provider can have the capacity to have every server utilize network 100%, but despite this the customer still benefits from increasing the uplink speed.
In my opinion, their limits are perfectly reasonable, the configuration is in the interest of the customer, and what is provided seems to be a reasonable interpretation fitting with existing definitions. It is only lacking documentation or unclear policy (one rarely plans for the unlikely, and I don't think they expected this), and something that should just be resolved by their customer service department.
Yes, if that's truly what the typical customer would want. But I doubt a typical customer would want only an ice cream cone after seeing that offer. However, if you offer "unlimited* ice cream", adding in the terms and conditions that it can't exceed 1 gallon per day, and 99.9% of users don't eat over 1 gallon per day, then I think it's fair to serve only 1 gallon per day.
Some years ago I purchased a (what I thought to be) huge backpack from an ebay store. it said 200 liters. What I got, I estimate to generously be around 75 liters. (by measuring the outer dimensions)
I complained, and they refused to budge, even after I showed them photos of the measurements, and they responded with "we have sold 3597 of these backpacks and nobody else complained about the volume, you must be wrong!!!" (i dont recall exactly if it was 3597, but it was thousands, and a specific number).
Clearly most people got what they expected, a reasonably large backpack, but not a HUUGGEE backpack. Was this okay? Was it acceptable I had to argue countless times and in the end contact paypal to get my money back?
Why is it okay because its hosting services?
edit: and it wasnt just misprint on the listing, the 200L was embroided into the backpack itself, and similarly on their smaller models, with the smaller number
If you really want to use this analogue, then the provider is getting upset after continuously providing 8000 gallons of ice cream every month to an individual who bought a single ice cream cone with "unlimited refills".
They are advertising that you can use 100% of your 1gbps link in a month then cutting TFA OP off at 80%.
If your interpretation of basic words is so wildly different from the norm that you already consider "1Gbps unlimited" a contradiction then surely you would never find yourself in a contract where subtler terms are in question.
1gbps maxed out for a month (you know, unlimited usage of a 1gbps line) equals 330TB (give or take depending on the length of the month).
If you signed up for a service which offered 50TB of upload then they cut you off at 35TB then you would be rightly annoyed.
If you have caps on your sevrvice? Cool. Don't use the word unlimited and inform customers of those limits.
Sure, I might be within my legal rights to drop you as a customer the next month, but isn't it still a scummy and misleading way to advertise my service?
Eventually, you either get an ISP-like situation where everyone has caps on their "unlimited" service and there is no room to disrupt the market, or you get MoviePass.
There is not going to be a free-market solution to this problem, because no service with defined limits will be able to compete with the "unlimited" service at the same price point, and there are no consequences to simply dropping the expensive customers.
Are you implying that this unintentional? That things like this are marketed without the full knowledge that such a service is impossible?
At a fast food places they advertise unlimited refills on the cup they sold you as part of the meal.
Here you are being advertised unlimited traffic on the port they sold you as part of the server.
Trying to liken using the port you are being sold as part of the server to driving up a tanker trunk to a fast food place is silly.
You will probably not stay in the establishment very long.
It is the normal operating mode of an internet connected server to send traffic to the internet - often even 100% of the time! It is the normal operating mode of drinks at a restaurant to be drank by the person while there, and maybe one refill for the road.
Trying to liken "Using the 'Unlimited' 1gbps port that was advertised "to "Refilling a cup and dumping it into a garbage bag" is ludicrous.
I don't understand why people are doing mental gymnastics to defend a bad advertising practice.
It is? I've been admining servers for a long time and I've never run one using 100% of it's upload bandwidth 100% of the time. I've never even run one at 80%. Being that near capacity means your service is about to fall over. Frankly for most services it's probably flagging.
So if a link is congested and a packet has to be dropped, then it is dropped from whichever user has sent the most packets in the last 30 days.
The end result is light users will get great service. Heavy users will start to see congestion at busy times, but will still be able to download as much as they like, contending with other heavy users at busy times.
Actual implementation can consist of say 10 QOS queues, where each customer is assigned a queue based on their decile of traffic sent in the last 30 days. Ideally you'd do strict priority on the queues, but I actually find it works better to say each queue has 2 packets delivered for each one of the queue below. That ensures that if a low use customer has a big burst of traffic, other clients still get a few packets delivered, which helps flow control adapt better. Also, rather than just ranking on 30 day traffic volumes, it works better to rank on 30 day traffic, plus 7 day traffic, plus 24 hour traffic, plus 1 hour traffic. That effectively means traffic sent in the last hour is counted 4x. That means users get faster feedback.
Do this and your company gets better utilization of network hardware too - every link will be 95+% full all the time. Thats better than the traditional model of building for peak usage and having links sit idle at 3am. There are plenty of customers who would love to use that cheap 3am capacity!
That's still overpriced by a factor of 10. Hetzner overage price for the 10G link is 1 euro / TB or about $0.001/GB. That seems pretty reasonable.
So we're really looking at something more like $250/month.
Likely in this case, they can't upgrade their network fast enough, I'm guessing. And they don't have the funds to do it.
This would be a different story than my local cableco say. They will sell you 5TB of data for $100/mo. Or about $0.02/GB. But that money isn't going into network upgrades. And won't until they're forced to, since that's the cable company business model and has been for decades.
I fully agree. People use more and more bandwidth. It was just a question of time before someone started to use a petabyte a month. "Unlimited" makes no sense unless they are really ready to allow 100% usage of 1Gbps uplink all the time.
Nah, even with fully saturating the link you won't get there.
I would personally assume that it would include 100% 24/7 (barring reasonable service outages). That's what "unlimited" means.
People who own dictionaries: this is a completely idiotic line of thinking. By which I mean "reasonable". Why, is there a problem?
You also can't set a max billing limit (or a soft limit).
It's not obvious that a script got caught in a loop and decided to write up a $5000 bill overnight.
Even so, I don't think it is AWS's fault that buggy code was left to run unattended overnight in the example you give.
It is absolutely 100% Amazon's decision to give you little to no tooling to prevent this in many scenarios other than going to their billing department and beg for a correction.
I do not use any service that can proceed to bankrupt me for personal use for what could be considered a normal use case. At work I'd pass the issue off to my manager who'd pass it off to a rep and deal with it that way, but there is no way I want to take that responsibility on myself to correct.
That aside, I also understand that they can't actually put a hard cap, since usage is measured after the fact occurs, and by the time the system triggers a halt, more usage has likely occurred. At the very least, I'd be fine with paying $X + 1 when I set an auto-shutdown threshold to $X if that was what they had to do to implement it.
Outside of that quibble, I find the other complaints to be less reasonable.
Back to your analogy. You get charge for these other dimensions that vary when you rent a car:
- make, model
- miles driven
- toll usage
- driver’s age
- duration of rental
The more accurate analogy here for the GP is the person returned a rental car to the wrong lot and was charged additional days until they discovered their error.
more like 5-6k max at a major IX point.
this also ignores how much of their traffic is settlement free peering to other adjacent ISPs.
IMHO Hertzer is the only hosting company that get what business it is in. AWS makes a ton of sense if you need to rent serious iron to do OCR on 100+ GB of scans of corporate documents, you need GPU power to train your AI model, or you are a startup that needs to grow 50% a month.
If you are an established company with a normal demand model, AWS will rip you alive.
Breakeven time between paying for GPUs, servers, and colo vs GCP was 3 months when I last checked. AWS was even worse. For training, cloud GPUs only make sense if you don’t have anyone on your team that’s familiar with physical infra or if you have a ton of free credits which is fairly common for startups with VC funding.
AWS is like a hotel. Great if you need it for a couple days, but you are mad if you want to live there indefinitely.
Although I bet $LARGE_AMERICAN_ISP will not warn you, but just shut down your site and lock you out of your data.
Do you live your day-to-day telling people lies or what?
None of that behavior sounds rational to me.
I can only imagine 'what? you thought I was going to paint the ENTIRE house!'
809Mbps costs $81/mo for a small provider.
There is in fact a very clearly stated limit. It's unlimited.
That would be $16,691 worth of transfer from AWS (US East pricing for 250TB to the internet). I'm not arguing that AWS isn't overcharging, just putting up a comparison. Even at $0.01/GB, that would be $2,560.
They should probably just say 50TB with overages at $0.005/GB or something like that. They are offering 20TB on their cloud servers so 50TB on the dedicated seems reasonable and $0.005/GB for overages is less than you'll get almost anywhere else while still probably enough for Hetzner to make a profit.