2. While I've always favored erring towards keeping people's sites up we are currently working on changing the default behavior to never let free sites incur overages
i.e. if I know I average $10 a day, I should be able to put in a "If it hits $50, email me and take it offline".
Of course the opposite problem is then people setting that limit too low but since the user defines the limit that's on them not you.
This is one of the reasons I still in 2024 rent physical boxes and run the modern stuff on top of them directly, yes it costs me more per month but the price is hard capped.
It's bizarre to me that more services don't support billing this way, since there are tons of situations where I would much rather have a site or service go down than be hit with a surprise bill and have to depend on social media and magnanimous corporate PR.
You can do stuff like "disallow any of these instances to be used in your env", so if you never use graphics cards, disallow the whole class.
You can also set limits like "no more than 20x m5.4xlarge".
But again, AWS is the worst about no actual hard limits, cause each system generates bills. Ive also seen the hell of "hidden system AWS Billing doesnt have is still submitting billing and we dont know what it is". Again, AWS enables basically infinite liability.
Ive also discussed with C levels that "every engineer and dev with AWS logins have an unlimited credit card to of which you're on the hook for". Lets just say that 'heartburn' doesnt even begin to describe the terror on their faces.
The big advantage to "cloud" was that you can provision more resources in seconds. Existing data centers had terrible non-VM and non-good VM management software that provisioning more CPU, RAM, storage was a weeks or months long trial.
And you can still get a 100k bill for hosting a SINGLE text file, cause DDoS and shitty hosting providers who demand unlimited credit is a thing.
That said, I still think there should be an easy way to set a daily limit. Should I use the Budget service to do that? Cost Explorer? Billing Alarms? Is it possible to have them shed whatever's spending all the money?...
Again, I see the whole can of worms here: what if your service is jamming tons of data into S3 because of a bug? Or you actually started something that got popular and you have a gigantic Dynamo table? Stopping an EC2 instance is maybe an easy call, but deleting data is iffy.
AWS just feels like a minefield because I'm occasionally worried with all the products, I'll check a box when creating an instance or SG or whatever, and that'll (e.g.) trigger CloudWatch to read all my logs, but I have some crappy debug config for some app which will vomit out dozens of logs a second accidentally, and instead of just trashing `/var/log/` I get billed for millions of log events or something.
They might even offer this as an insurance, so you pay a little more but can be sure you stay in budget.
I stopped ~all autopays when (Boost Mobile?) deduced 200 instead of 20usd from my debit account one month in college. They refunded the difference relatively quickly, but I racked up 5 or 6 overcharge fees before realizing; ended up being a pita for an already broke 18 y/o to figure out.
I was a broke and stupid kid. Never use a debit as an autopay. But since then I like to track where each penny goes as it goes.
How much is unlimited talk, text, data, hotspot with a watch data add-on where you are from?
edit: Actually I was wrong–it is $30, not $35.
I could also get 50 MB data, 2h calls, unlimited texts for 2€.
I pay $20 for the phone plan, and $10 for the watch add-on. No fees or taxes Feels reasonable to me.
What Netlify is doing here is really the best approach for both parties. And typically speaking a $104k bill would be hard to get paid up regardless if the customer's typical transaction balance was $5/mo and their credit card limit wouldn't be that high.
Also, that's the benefits of credit cards - that you can still issue a charge back, and credit card companies very much favor the consumer rather than the merchant.
So your suggestion is to issue a chargeback.. to get money back that should under the terms of whatever service you signed up for be owed?.
That seems like bordering on fraud tbh.
> Additionally, when you set that limit people then get upset because usually when they go over it for a good reason, like going viral, they aren't anticipating it, and just when their traffic is most valuable the site is down.
Legit concern and something I mentioned, I'm gonna guess there are broad two camps on that one - mine which is "I want a safety ripcord" and "whee, nice problem to have".
However since this entire conversation is around a guy who got a massive invoice because of a bill he wasn't expecting and couldn't have set such a limit I'm still gonna go with a "I want a way to constrain the financial downside - hell turn it off by default but give me the option".
Since broadly a lot of cloud stuff doesn't, I'll constrain it a different way.
Funny story... One of the big cloud providers actually has you do that on purpose as a remedy for an account you've lost access to.
That has not been my experience. I've had to do a few chargebacks for services not rendered, and I've never won. I will submit my evidence, then the vendor will submit 100 pages of random emails, and then I will have my claim denied. Then I will appeal, will point out that they sent 100 random pages of email, and then they will reply with the same 100 pages of emails and I'll get denied again.
It seems that the vendors have found the hack for chargebacks -- just inundate the credit card company with so much data that they assume the vendor must be right.
It makes sense -- the vendors pay the credit card companies a lot more than I do. They'd rather keep them happy than me.
Plenty of morons just use the service and chargeback right before renewal so they got the service for free, some just chargeback instead of cancelling their plan or asking a refund. I get hit with 15$ bill plus I lose all the money even if I provided the service.
Whatever email / data from my system I send is ignored and the scammer / moron gets his money back.
I'm sure that's how it works if the vendor is large enough.
If you are a small fish you don't stand a chance, which is why for my next project (where I'm supposed to charge a lot and spend a lot on behalf of the user - and I'll be royally screwed if I start getting chargebacks on 500$ of which I've already spent 450$) I'll just accept crypto payments.
Additionally, when you set that limit people then get upset because usually when they go over it for a good reason, like going viral, they aren't anticipating it, and just when their traffic is most valuable the site is down.
Instead of forcing user to set a low cost limit and missing a viral opportunity or the platform writing off the massive bill the customer can't afford... just put the billing mode into a reduced price mode or have some more nuanced configurations. Sometimes is just asking the question the right way. Instead of "max spend limit" or similar "If your site goes viral, how many requests do you want to serve before going offline? 1M=$20, 10M=$100, etc" at this point, I feel like bandwidth consumption is a bad metric for billing; just use requests/visits/actions and price for those.
This is not prescriptive just illustrative. The point is make a better pricing formula to account for these massively unexpected events. Couple it with an aggressive notification policy when this traffic event gets triggered. The user should know the traffic pattern has changed and a high traffic event is happening. They can login and change the configs and decide if they want to keep it going or not.
I agree. I also agree that when dealing with large numbers of people, there will be people who don't understand this and/or will actively try to social engineer their way out of their own decisions.
Setting customer expectations and meeting them successfully isn't easy.
The fact that most cloud operators don't have actual hard cutoffs to maintain financial responsibility is intentional. Azure does, but only for specific account types. If it's PAYG, you can't do it. The end result is if you do something "weird", or someone DDoS's you, you're liable.
With a hard limit, a DDoS just takes your site offline.
There would be many attorneys interested in collecting a $105k debt.
I mean I'm not familiar with every debt collection scenario under the sun but Internet randos seem to think this is a real thing where like a cloud/hosting company sends an army of lawyers to repo some guy's house and runs him into bankruptcy because of a traffic overage. I've never seen it work that way, what happens like with most business debts, is someone at the company negotiates with the debtor to try and get as much out of them as they can, and failing that, possibly refers it to a collections agency which does the same but plays a bit more hardball.
In the case here with Netlify even before it went viral they reduced the amount from $104K to $5K, no lawyers, collectors or repo men involved, and while I'd hate to be stuck with that $5K bill, I dunno, that does feel closer to the mark of something that maybe you should be on the hook for if you're responsible for 200 TB of bandwidth overage over 4 days? Is this so bad on the part of Netlify?
All that said I'll just add that I've never given my credit card to any sort of host/cloud who had terms where they could bill unlimited overage fees like this. Never will unless there's a cap. Not Netlify not AWS not nobody. That goes for my personal life as well as for the business I operate. The terms is the terms and the answer is to not use these services unless you can afford them imho.
The responsibility part is the tricky part of the equation.
If someone hits your site with a DDoS attack, are you responsible? There's literally nothing[0] you can do as a customer of a cloud provider here because anything you can do is limited to the servers and services you're given access to. For example even if I had access to billions of requests and built an anti-DDoS tool it would still need to run within the cloud provider's provisioned server which means I'd be on the hook for all traffic costs because it's something running in my account.
That doesn't seem reasonable to me as a customer. It means a cloud hosting provider can put an extreme financial burden on a customer and make a killing in profits because of the markup they charge on bandwidth. The incentives are terribly misaligned.
[0]: I mean you can sign up for DDoS protection through a 3rd party company but in this case I'm talking about taking actions within your hosting provider.
From personal experience as a customer of a cloud provider (not with Netlify btw), usually cloud providers who profit from bandwidth costs will write their TOS in such a way where almost nothing qualifies as a DDoS attack unless it's truly a distributed and targeted large scale attack specifically on your site.
A random person on the internet who spins up a few VPSs around the world and slams your site with looped curl requests won't count as a DDoS attack even though from your perspective that will result in a massive bill increase due to bandwidth costs.
In other words, I'm not surprised "didn't match attack patterns" was used. I'm guessing that will be the case most of the time.
Mostly false. Either transit is billed on a 95th%ile basis (so...more money for more traffic), or if it is flat/netted, you're still paying for the capex for the switch ports (fatter connection to support more traffic means more $$$ for the gear to support it).
You can say "the sum total of free-tier traffic requires X additional connections" but the sudden burst of DoS traffic did not raise marginal costs (besides an inconsequential usage of electricity).
If they're about to go over, shoot them a quick heads-up and give them 24 hours to sort things out or level up their package.
After that, if they haven't made any changes, temporarily pause their site access
By the way, your comment was flagged which seemed odd, so I looked at your profile and it seems like all of your comments are being (automatically?) flagged and don't appear on HN by default. You might want to talk to the HN staff about that.
2. To me, it's more likely they don't have one because there are edge cases to consider that make "hard limits" difficult to implement. What is AWS supposed to do when you hit the limit? Is it a hard limit? Ok, so when I hit my budget, all my s3 buckets get deleted and all my EBS drives get dropped? Do all my code deployments just get deleted? Do you "bless" certain services so that they continue to charge the user even after the hard limit? How is all of this communicated?
You can set an alarm in AWS today and the user can decide what to shut off. If you really need to, you can create a script that can hard nuke your account once a limit is reached; but I don't see why AWS should nuke your account for you.
If providing this bandwidth cost them $50k rather than $5, I would bet you my entire life savings that they would QUICKLY find a way to add hard limits to their service, no matter what technical challenges they're quoting now.
2. This is probably a 95/5 problem. The vast majority of these sorts of cases that I've seen and read about are caused by increased traffic, which hits the customer either with extortionate egress fees or unintended compute scaling behavior (FaaS/VMs).
Storage is trickier, but you could stop accepting writes or reads after passing some hard limit.
This leaves some edge cases, sure, but it should handle the vast majority of unexpected bills without any destructive actions.
> You can set an alarm in AWS today and the user can decide what to shut off.
That shifts responsibility back on the customer, which is exactly the problem to begin with. I need assurances that I won't be billed more than $X in any given day, or month and this solution doesn't provide that. Maybe their API down, maybe it's serving stale data, or maybe my automation fails for some reason.
> How is all of this communicated?
Hard monthly limits:
Egress: [ $ 10 ] - When exceeded, all outbound traffic is limited to [ 0 Mbps ].
Compute: [ $ 10 ] - When exceeded, scale all compute instances to [ 0 ]
Data Storage: [ 1 TB ] - When exceeded, all writes are rejected
Reads Ops: [ $ 1 ] - When exceeded, all reads ops are rejected
It's really not that hard. Offer these limits on a per-project, or even per-resource level, that would naturally allow you to limit the blast radius. A personal website that has gone viral would likely want to have different limits than an email server under the same account, for example.
No, I believe you are assuming that proving a "hard limit" feature is cheaper than just refunding people from time to time. If you consider the all the products AWS has to offer, all the different ways they are billed, then figuring out what do to for each product once that limit is reached, and potentially doing a destructive action, and then all the code and testing on top of that - it seems far easier to add a human in the loop to just refund people on a case by case basis. It's a ton of complexity for likely a rare issue, and if someone really needs it they can build one themselves and choose how to handle what needs to be done once the limit is reached.
Building all this out just so that some college kid isn't charged a bill they could obviously never pay likely isn't a good use of resources. It's not likely AWS has a reputation of being greedy either, if you explain the situation they refund you. If they are truly doing this to make money they have remarkably poor execution.
>It's really not that hard.
Storage pricing isn't done up front. You pay per gigabyte-month. If your "hard limit" kicks in, then what does Amazon do with the data you are no longer paying to store? Does Amazon drop your entire database once the credit limit is reached? There are plenty of Amazon services like this. Consider secret manager. You are charged 40 cents/month per secret. You have 100 secrets, so $40/mo, but you set your hard limit to $20. The middle of the month rolls around, what does Amazon do? Do they just drop all your encryption keys?
There are 100s of AWS services and many where you can't just apply a sensible rejection policy once you hit that hard limit without doing something the user cannot recover from. It's only not hard because you aren't actually thinking about the matrix of AWS products that exist and how they are billed.
I don't think you understand what a hard limit is. When I set a hard limit, that means that under no circumstance should I be billed more than this amount. From this you should be able to deduce that the platform needs to stop you from ever entering into a situation where they could end up having to decide between nuking your data, or overcharging you.
This means that for all data storage scenarios, they should apply limits to projected monthly usage rather than actual monthly usage.
If 1 GB of storage costs $1 per month, and my monthly limit is set to $10, then they should not allow me store more than 10GB of data, rejecting writes as needed. This design guarantees that my hard limit won't be exceeded and the data is never at risk.
Any service which doesn't persist data can apply limits based on actual usage, and safely shut them down when the limit is exceeded.
> Does Amazon drop your entire database once the credit limit is reached?
No, they should refuse further writes when your database reaches a size that would exceed $10 a month. That's a safe behavior.
> Consider secret manager. You are charged 40 cents/month per secret. You have 100 secrets, so $40/mo, but you set your hard limit to $20.
If you set your hard limit to $20, then Amazon should refuse to create more than 50 secrets. That's a safe behavior which limits your spend to at most $20.
> Building all this out just so that some college kid isn't charged a bill they could obviously never pay likely isn't a good use of resources. It's not likely AWS has a reputation of being greedy either, if you explain the situation they refund you. If they are truly doing this to make money they have remarkably poor execution.
This is just a bad faith argument, everyone from broke college students to medium sized businesses has complained about this at one point or another. Relying on the good will of a megacorporation to forgive a surprise $100k bill is nuts and if you scroll through these comments you'll find some that claim that AWS is no longer forgiving these bills like they used to.
I'm not claiming that this solution is perfect, or that it would cover every failure mode, but it sure beats the status quo and would solve the vast majority of surprise bills that result from unexpected traffic.
>No, they should refuse further writes when your database reaches a size that would exceed $10 a month. That's a safe behavior.
I think we have gotten to the point where we've lost the plot here. You previously stated "It's not that hard". We are doing projection forecasting! And the storage layer for these range of services (S3, MySQL, Postgres, MSSQL, Kafka, Keyspaces, DynamoDB, probably 20 more im missing.) must now either call out to some financial projections service and reject writes for billing reasons. On top of that if you have a hard limit you are locked out of dynamic scaling. Need to 100 nodes for an hour? Nope, you now must disable billing limits because Amazon assumes you will use every resource for the entire month. Surely someone wont disable billing limits for an hour and forget to re-enable them, right?
>but it sure beats the status quo
I'm not sure handicapping every service for some barely functional billing limit beats the status quo. It's certainly more expensive for amazon in terms of engineering load but with so many footguns I don't see how it ends up anyway other than Amazon having this feature and still having to do service refunds because billing limits were disabled due to awkward limitations.
>This is just a bad faith argument, everyone from broke college students to medium sized businesses has complained about this at one point or another. Relying on the good will of a megacorporation to forgive a surprise $100k bill is nuts and if you scroll through these comments you'll find some that claim that AWS is no longer forgiving these bills like they used to.
Medium sized businesses have had their bills forgiven for things like cryptomining for example. My point is billing limits don't really work for AWS, alerts is the best you will get because AWS doing destructive actions for you or assuming your usage patterns is worse. It's not an easy feature and to imply they keep the status quo solely because Bezos needs another yacht is reductive of the problem.
It's not a complex "forecast". It's simply the monthly rate as I've illustrated with numerous examples.
If my limit is $10, then don't let me occupy more storage than that. It's a 1:1 conversation between gigabytes and how much they cost on a monthly basis.
Do you have ties to cloud providers you'd like to disclose? It's a very simple concept that you're refusing to understand, which usually happens when your job depends on it.
I have 100 secrets and no hard limit. Now I set the hard limit to $20. What does AWS do?
Also: what's common? At the scale these companies so desire, small percentages are still thousands of users. Supporting them with what's ultimately a fairly trivial option shouldn't be seen as such a hassle.
1) Thank you for allowing us to set the limit.
2) I understand your opinion that you prefer chargebacks but I disagree with it.
The very reason I stay with Hetzner is that I know in advance what my bills will be for the whole year. Heck, I even charge my account in advance so that I don't worry about any charges!
Most users don't want to be banned from their hosting provider.
Of course. And that’s why any limit against a dynamic variable should also have alerts linked to it
Send an alert to the user when traffic starts spiking, especially if a simple projection shows it’s going to go over their limit
Then the user is aware, hopefully with enough time to lift the limit if needed
If your customer is aware enough to notice they are being hit with a DOS or legit traffic while it is happening, then great! They can respond, and if needed, engage proserve to get support for scaling or defense depending on needs.
If your customer is not alert enough, then their site is offline, and they won't hear about it until their customers are screaming at them, which will result in a P1 ticket to look at a vendor who won't turn them off during an unexpected peak.
It's a catch 22, and if you have to choose between: a) a PR hit because you have to go on a forum and post about waiving the fee, or b) a PR hit because someone posted a blog post about how you killed their site during a moment of critical growth
any reasonable business will choose A every time because A is far more supportive of customer growth and has drastically better optics. Anyone who thinks A is worse is probably too inexperienced to have an opinion.
I'd say the same about those who only think in absolutes.
That's a false dilemma. You can give your customers a choice in these matters with an optional hard limit, which I realize seems like a rather extreme idea these days.
Someone running a personal project will likely opt to have a low hard limit, say $10, while businesses will more likely opt in for alerts without, or with very high hard limits.
Sort of, but the approach to the approach isn't great. If you're going to void charges from DDOS traffic anyway, you might as well make that explicit policy, rather than doing it after the fact in a way that seems discretionary.
The Reddit thread is full of people who are saying that they intend to pull their static content off of Netlify and move it to Cloudflare Pages, which has no overages on the free tier in the first place. I can tell you that I personally chose Cloudflare Pages to set up a new static site a couple of weeks ago, and Netlify wasn't even in the running.
If Netlify's free tier is important to their customer acquisition strategy, then they really need to retool how they're offering it, or Cloudflare is going to eat their lunch.
Hard limit: [$1000]/mth or [$500]/24hr period
Notify me if traffic exceeds thresholds: [$800+]/mth or [$400]/24 hr period
Notify me if the traffic forecast for the month looks like it will exceed my hard limit before the end of the month.
Like let them go over one month and then make them sign up for a paid plan for the next.
I still prefer this too. Kinda funny how server resource limitations became a feature and not a bug when it was one of the problems the cloud sought to overcome
>We normally discount these kinds of attacks to about 20% of the cost, which would make your new bill $20,900. I've currently reduced it to about 5%, which is $5,225.
20% and 5% are quite a bit higher than forgiven.
With all that money they then can finance the free tier until they get too far and become platform locked-in.
Surge.sh Im not sure. But shows all the sign of some greedy acquisition, regular long outages , as if I have been sitting as a free tier for too long, quick nudge to pay. For barely accessed sites even behind CDNs, steep. I even worry they one day just wipe all my buckets (they did for a few already) and support would recommend me to be a "normal" paying user .
Nothing is free. And nothing too good to be true is true .
I wouldn't want to be CEO these days. A lot are trained and paid to do damage control.
This isn't what you said in your first post, you said:
> It's currently our policy to not shut down free sites during traffic spikes that doesn't match attack patterns, but instead forgiving any bills from legitimate mistakes after the fact.
So forgiving "lots and lots" doesn't move the needle. Do you or do you not forgive _all_ such cases where your DDOS protection doesn't take down the site? What was your employee referring to when saying that the usual discount is 20%? Are you saying that you _never_ discount 20% and instead always discount 100% i.e. "forgive"?
Hey I'm a taxi driver. Hailer fell asleep on the back, so I kept driving all night, once he woke up I dropped him to his place and asked for my monthly wage. I "forgive" many, but just a few are juicy income so I adopted the policy to never wake any customer up. If people ask I say it would be impolite, principles prime.
Sequence of events doesn't support this answer:
1. User gets charged 100k
2. User complains to support
3. User receives discount to 20k, then 5k. Support states policy is normally 20k
4. User discloses to the world. Goes viral.
5. Invoice is forgiven
While you might forgive "lots and lots", fact is that you still presented the invoice to a free tier customer, and when they complained you gave them a discount, but still charge. Only when it went viral did you forgive it.
I'm a paranoid person by nature so "It's free... just... give us your card details" is always suspicious.
1. What are you doing to prevent DDoS's from hitting your network?
2. Why do customers have to allow an unlimited credit burden to use services?
3. Why arent there cost controls to "if $$ exceeds X, shut acct down"? Azure can do this.
Long story short, why are you by default (except for social media escalation) passing fraud costs to customers?
It seems that significant bills appearing without warning or cut-offs is clearly intentional. I am embarrassed that I recommended Netlify before.
Solving this only for the free site use case doesn't address the core problem that people are bringing up about a lack of spending limit controls.
From what I see you could change this policy tomorrow unilaterally and we would have no recourse.
I wouldn't think it's a binding policy at all, because the billing procedure (automatic full bill, manually discounted bill, etc.) would follow it if it were. More of a procedure.
No offence, but this sounds like "trust me bro" billing and it is not good enough. Someone could literally get a heart attack from getting $100,000 bill - this amount of debt can literally ruin someone financially.
> 2. While I've always favored erring towards keeping people's sites up we are currently working on changing the default behavior to never let free sites incur overages
I hope you understand that chance someone who used to pay you $20 / month unlikely want to ever get $10,000 bill. Yeah people might dislike that their website went down due high traffic, but it's not gonna bring this much negative PR as incidents like this. There should be some sanity check at least.
Not a heart attack but there has literally been at least one suicide over an unexpected life-ruining bill like that:
https://www.forbes.com/sites/sergeiklebnikov/2020/06/17/20-y...
But this is a good reminder why my gut feeling always made me avoid these overengineered solutions.
They do. But of course maximizing profit is the sole true prerogative of capitalist enterprises. And the market is not totally competitive. So yes your intuition was correct, to be cautious against over enginnered pricing to get y'a.
I mean those companies cater to hobbyist. Then ...
Render seems more fair-play. Until a change of mgt occurs of course.
That’s the elephant in the room here. I understand an enterprise plan where you state billing is $xx per GB, but billing someone with a free site??
Give me a break.
Honestly, this terrifies me---I run a bunch of different sites off netlify, and I would have never imagined that a site could jump from 0 to six figures of bills a month without something hitting a tripwire somewhere and cutting it off or at least communicating with the account owner. At least users should have the capacity to self-impose bandwidth caps to prevent this sort of thing.