I got pwned by my cloud costs

Don&#x27;t put Cloudflare in front of a Cloud egress bill. i.e. don&#x27;t do this: Azure|Amazon &gt; Cloudflare<p>Always use your own proxy where the egress is well within your free tier, i.e. do this: Azure|Amazon &gt; Hetzner|Linode &gt; Cloudflare<p>Why?<p>Because Cloudflare cache is a massively multi-tenant LRU cache and whilst hot files will be cached well (and with Cloudflare Tiered Cache even better - but this itself is a cost) anything else is still going to expose you to some degree of egress cost.<p>When I exposed AWS to the web I paid $3k per month to AWS. With Cloudflare in front of AWS I paid $300 per month to AWS. With Linode in front of AWS and behind Cloudflare I paid $20 per month to Linode and about $12 per month to AWS.<p>A Linode, Hetzner instance... or any other dumb cheap web server that comes with a healthy free tier of bandwidth is all you need to set up a simple nginx reverse proxy and have it cache things to disk <a href="https:&#x2F;&#x2F;docs.nginx.com&#x2F;nginx&#x2F;admin-guide&#x2F;content-cache&#x2F;content-caching&#x2F;" rel="nofollow">https:&#x2F;&#x2F;docs.nginx.com&#x2F;nginx&#x2F;admin-guide&#x2F;content-cache&#x2F;conte...</a>

Ah the old cloud provider switcheroo. Yip this is the way they make money. They make it easy to setup some gigantic hugely scalable website then hit you with a gigantic scaled up bill. AWS would do this as well.<p>Team I&#x27;m in at the moment is in the early stages of cloud adoption but the company in total has fell hook line and sinker for AWS. When I mentioned the cost there is always an excuse.<p>The main one being that you don&#x27;t have to hire sysadmins anymore as that&#x27;s taken care now by AWS. Ah yes but they have actually been replaced with a &quot;DevOps&quot; team plus just our department now spend &gt; £1 million per year to AWS in hosting costs. A 20% reduction in those fees could pay for a few sysadmin(s).<p>The next one is that no other vendor would be able to supply the kit. You know StackOverflow is able to run on a single webserver (<a href="https:&#x2F;&#x2F;nickcraver.com&#x2F;blog&#x2F;2016&#x2F;02&#x2F;17&#x2F;stack-overflow-the-architecture-2016-edition&#x2F;" rel="nofollow">https:&#x2F;&#x2F;nickcraver.com&#x2F;blog&#x2F;2016&#x2F;02&#x2F;17&#x2F;stack-overflow-the-ar...</a>). Plus many of the other providers have loads of instances available.<p>I mean I&#x27;m not against cloud it&#x27;s just not the cheapest option if you choose one of the big 3 providers. I use a company called scaleway (<a href="https:&#x2F;&#x2F;www.scaleway.com&#x2F;en&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.scaleway.com&#x2F;en&#x2F;</a>) they have all the essential cloud services you need and everything else you can run yourself in docker or k8s.

Ouch. If Troy Hunt of all people can make this mistake, it can happen to anybody. HIBP is an awesome service funded totally by donations, so it&#x27;s too bad this happened. Of course Microsoft is happy to hide behind their confusing pricing model and let customers overpay for Azure without alerting them.

It is worth mentioning that the alert itself costs money. So if you&#x27;re evaluating the alert every 5 minutes on the past 24h of data it can burn a small but surprising amount of money.<p>From TFA it looks like that would be 10 cents per &quot;time series&quot;. Or what I translate it to, is 10 cents every 5 minutes (*I think, but I havent used Azure in some time*). $1.20&#x2F;hour, $28.80&#x2F;day, almost $900&#x2F;month. Not too hard to drop that by making the alert less frequent. (edit: I think I saw AU$ there, so maybe it is AU$900.)

These stories almost always boil down to this fundamental conflict of what you want for a personal project vs a business. (though in this case yes, Troy Hunt&#x27;s HIBP is larger than a lot of startup businesses)<p>In a business setting, you want your service to stay up, at the cost of spike in costs if accidents or mistakes happen.<p>In a personal project, you want there to be hard limit on cost, and your service to go down if spikes call for it. (I&#x27;m relatively sure that no one wants their personal projects to incur a bill of thousands of dollars by accident.)

Have you contacted Azure? On one hand you owe the money “fair and square”, but on the other if I were them I’d waive an unexpected $10k bill to a good faith actor that was incurred without any proactive notification by Azure.

I wonder if before cloud computing, has there ever been a successful product &#x2F; service where it was accepted with just a shrug that the volatility of monthly costs means it could bankrupt you with next month&#x27;s bill, because of complexities and opaqueness of the cost structure make it virtually impossible to predict and protect against extreme peaks in all parts of the setup.<p>Even if you run a relatively opaque cost structure business like a restaurant, you can still calculate the maximum cost of ingredients for one month, the salaries, energy, etc. if you simply use the &quot;best case scenario&quot; of having every seat at every table booked for all opening hours, with people ordering your most sold dishes. Cloud computing is still leagues above that in terms of cost predictability.<p>I once worked for small, non-startup software company who pondered moving servers to Azure. The Azure partner shop analysed the needs and came up with a monthly cost &quot;between 30k and 120k per month&quot;. They were really surprised the company stuck with their non-cloud setup because &quot;everybody is moving into the cloud!!&quot;

That&#x27;s the typical story. Something goes wrong and it costs you (typically a small company) a lot of money. At that time just nobody is looking at metrics. Even alarms don&#x27;t help absolutely because they can also be missed.<p>The only thing that would really help were a hard spending limit that stops all services except storage. If your site is important there will be such an amount of user feedback that it is impossible to miss it for a long time.

Most worrying is that even an expert like Troy Hunt was UNABLE to figure out the cause of the issue by himself. He &quot;reached out to a friend at Cloudflare&quot; who investigated and found the cause.

Cloud providers should always have a max spend and it should be a standard feature. The cap shouldn&#x27;t even be some optional feature or notification service. It should be a hard cap that you can move - at your own risk.

One thing I hate about the cloud providers is that there isn&#x27;t an option to set a maximum cost. I would prefer to plug the cable of my side project than just receive an email saying me that next bill is going to be over my cost. I understand not everyone would like to do that, but I would like to have that option.

It would be really classy if MS forgave that debt, especially considering the service is a public benefit.

I wonder how much of the cloud provider revenue comes from situations like this. I suspect quite a lot.<p>I think that the cloud provider business model that allows for uncapped maximum costs is a bit of a commercial dark pattern. What makes it somewhat more nefarious is that it is relatively easy to blame the customer.<p>I’m not surprised that the cloud providers are quick to refund users as it’s likely that they only do it in a fraction of cases and it buys a lot of goodwill.<p>It would be interesting to try and design a cloud that supports OutOfMoneyException’s with gradual degradation and capped liability for costs built in.

Question, is the 0.014AUD per GB quoted here correct? Looking at the linked page[1] I would think the cost would be 0.1102AUD per GB as is quoted in the Internet egress section.<p><a href="https:&#x2F;&#x2F;azure.microsoft.com&#x2F;en-au&#x2F;pricing&#x2F;details&#x2F;bandwidth&#x2F;" rel="nofollow">https:&#x2F;&#x2F;azure.microsoft.com&#x2F;en-au&#x2F;pricing&#x2F;details&#x2F;bandwidth&#x2F;</a>

Interestingly, Troy says that egress is expensive on Azure at $0.014 AUD&#x2F;gB (~$0.010 USD&#x2F;gB), but that is the same price as additional egress for Linode and DO, and Linode egress has never struck me as expensive. In fact, I’m kind of shocked (as an AWS user) that Azure egress is the same price as Linode.<p>Actually, wow it seems AWS is also the same price as Linode and DO for egress. While Linodes and DO do come with decent free bandwidth, this is a surprise to me.

Reminds me of a time, we had a new site that was going to run on GCP, we had been using a couple co-located servers for years.<p>When everything was moved to production, URL went live, nobody ever did any kind of bandwidth checking, caching, no CDN, no cost tracking. $10,000 in our first week. That&#x27;s about 1&#x2F;4 what our total spend on the co-located servers was for the whole year. Boss flipped his lid and wanted to kill the new guy who was on the project.<p>After about 2 years we got rid of all the co-located stuff and were spending about 1.5x, but we had more apps, they served heavier pages, etc.

It is very good these things are getting publicized. More and more people realize these payment schemes for what they are: a scam. Every cloud provider that refuse to put a hard spending limit participates in this.<p>It is important to remember that not all cloud providers participate in it. For example, in Hetzner Cloud, they explicitly provide the maximum amount you are going to pay for a given instance or service in a given month. You are guaranteed not to pay more. Everybody knows why Amazon etc. refuses to do it this way.

My (naive) solution. Every new account by default has an SMS alert that trips at $100. It says<p>&quot;Your account has exceed $100 spend. Reply &#x27;SHUTDOWN&#x27; to shutdown all services, &#x27;STOP ALERTS&#x27; to never see this alert again, or &#x27;DOUBLE TRIGGER&#x27; to double the alert trigger value to $200.&quot;<p>$100 is arbitrary, it could be any nominal sum. The idea being that the user can double the alert each time they get it just from SMS. I bet 95% of users would double their alert limit to a comfortable point. The other ~5% will be power users who customize their alerts.<p>The idea that these companies couldn&#x27;t know what limits customers want is kinda silly. We can use the same techniques for alerts that we use in algorithms for expanding vector storage, for example. We can &quot;amortize&quot; alerts, so to speak.

Very nice writeup, thanks to the author for writing it so clearly for someone who is not familiar with the nitty-gritty to be able to follow it.

Shameless plug - the core of my work is about ensuring these unexpected costs never happen.<p>We have some recent case studies where we&#x27;ve successfully reduced cloud costs by 95%<p><a href="https:&#x2F;&#x2F;www.cloudexpat.com&#x2F;case-studies&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.cloudexpat.com&#x2F;case-studies&#x2F;</a><p>hi(at)cloudexpat.com - happy to help!

As soon as I saw &quot;17GB file&quot; i thought &quot;that&#x27;s what torrents are for&quot;. Otherwise one mistake and... Well this happens.<p>Or someone maliciously bypasses CF cache e.g. by parameters.<p>Cloud just is not suitable for any kind of volume egress. It&#x27;s a death trap. Like going on vacation with data roaming enabled.

I have 10gbits internet at home. Sometimes I wonder how many services&#x2F;people I could bankrupt by using it harder. Not that I want this, but more like, why is it even possible?

&gt; I have been, and still remain, a massive proponent of &quot;the cloud&quot;.<p>Mice cried and stung themselves, but kept eating the cactus.

This particular problem basically boils down to &quot;CDN providers don&#x27;t like caching large files&quot;, which is a very common problem. Everything else was configured and setup exactly right to not have a large bill.<p>Most CDN providers have a lot of machines out on the edges of their networks, and it&#x27;s understandable that they don&#x27;t stuff these machines with large disks, likely preferring smaller faster SSDs. But this is a very common pitfall of CDNs that needs more attention, along with messaging on the dashboards and settings pages.<p>I&#x27;ve had problems with no warning on Cloudfront, Cloudflare, Bunny.net all from not realising that my files were beyond the CDN&#x27;s cache size limit, but none of them seem to do a good job at surfacing this other than &quot;talk to customer support&quot;.<p>Cloudfront does list the max size clearly in the limits and quotas page, though, and if you front your S3 bucket with Cloudfront, you could turn caching off and still get the discounted bandwidth out rates (S3 -&gt; Cloudfront is always free, even if the file is fetched every time).

This is why I use fixed price offerings for personal projects.<p>A large bill is probably chump change for someone like Troy, for others it&#x27;s a year or two of savings. The risk is not worth it.

&gt;What we&#x27;re talking about here is egress bandwidth for data being sent out of Microsoft&#x27;s Azure infrastructure (priced at AU$0.014 per GB).<p>AUD $0.014 is roughly USD $0.01. Which I thought was reasonable. But on [1] only &quot;Data transfer between Availability Zones(Egress and Ingress)&quot; cost $0.01. Do transferring from Azure to CF count as that? Other Internet egress (routed via Routing preference transit ISP network) starts at <i>$0.08</i><p>I hope someone from Azure CS could give him a custom discount.<p>It is also worth thinking, the cost HIBP saved on Cloud &#x2F; Serverless over the years could have wiped out ( if not more ) by this single incident.<p>[1] <a href="https:&#x2F;&#x2F;azure.microsoft.com&#x2F;en-au&#x2F;pricing&#x2F;details&#x2F;bandwidth&#x2F;#faq" rel="nofollow">https:&#x2F;&#x2F;azure.microsoft.com&#x2F;en-au&#x2F;pricing&#x2F;details&#x2F;bandwidth&#x2F;...</a>

We&#x27;ve been thinking about this for a while, and if there is any way we can catch these types of cost spikes before they happen. We&#x27;ve managed to do it for Terraform resources using an estimation approach, and using a usage file, you can model expected usage-based resources (<a href="https:&#x2F;&#x2F;github.com&#x2F;infracost&#x2F;infracost&#x2F;blob&#x2F;master&#x2F;infracost-usage-example.yml" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;infracost&#x2F;infracost&#x2F;blob&#x2F;master&#x2F;infracost...</a>), but this one has got us thinking more about policies.<p>To be clear - we would not have been able to catch this one right now :&#x27;(<p>Would love to hear thoughts &#x2F; brainstorm ideas - is there any way we can proactively catch these types of cost spikes?

One wonders how Cloudflare can essentially absorb all bandwidth costs. But AWS and Azure are using them as a profit center.

Well, the cloud is just a convenient way of accessing someone else&#x27;s server.<p>Convenience always costs money, there is no (big) cloud provider doing it out of their own pocket or rather not optimizing for huge profits.<p>It&#x27;s the same as with any other service, really. So I don&#x27;t understand, why some people assume it would be different here.<p>(Note: I am not saying that Troy Hunt assumed this, but I know people who go to the cloud because &quot;It&#x27;s cheaper&quot;. It was never cheaper, on no project I worked on. It was more convenient, but in the end it was more expensive mostly)

I would be surprised if Azure doesn&#x27;t waive or reduce this bill dramatically. Something similar happened to me with AWS. I had a simple file upload service where files would expire if they hadn&#x27;t been accessed in 24 hours. Someone started using it to upload music and videos. I ended up with a high bandwidth bill on Amazon S3. I reached out and explained what happened, they waived the costs entirely (to the tune of $5000).

Valuable investigation steps to find the erring cloud resource, But as Troy concludes &#x27;Budget Alerts&#x27; would have saved him from this issue.<p>No matter what the traffic is, The first thing to do with any cloud service provider is to set the budget alerts according to our wallet, be it one with credits or otherwise. At this point, I don&#x27;t even try any new cloud service provider who doesn&#x27;t offer credible budget alerts.<p>Another key takeaway is,<p>&gt; Huh, no &quot;CacheControl&quot; value. But there wasn&#x27;t one on any of the previous zip files either and the Cloudflare page rule above should be overriding anything here by virtue of the edge cache TTL setting anyway.<p>Even this could blow up. All cloud service providers set the &quot;CacheControl&quot; to &quot;No&quot; and if we would want to cache something which is not cached by CF by default e.g. *html using Page Rules then we need to set CacheControl (e.g. max-age) at the cloud service provider end too.<p>P.S. I&#x27;ve written about these recently on my blog titled &#x27;Saving Cloud Costs&#x27;[1] from a frugal solopreneur PoV.<p>[1] <a href="https:&#x2F;&#x2F;hitstartup.com&#x2F;saving-cloud-costs&#x2F;" rel="nofollow">https:&#x2F;&#x2F;hitstartup.com&#x2F;saving-cloud-costs&#x2F;</a>

This is why I personally won&#x27;t run projects on infrastructure with what roughly equates to unlimited risk billing.<p>It&#x27;s my opinion that it&#x27;s better to work with known limitations and optimize for them.<p>In the case of bandwidth, work with a fixed pipe size, or do the math and set up a QoS that implements a throttle to avoid exceeding your bandwidth allotment.

First thing one should always set on any cloud account is billing alerts. Set &gt; 1 and set first to ~ 80% of what you think will be your normal cost then add extra alerts all the way up to 100%. That way you&#x27;ll usually get an early warning with some time to act before it becomes really expensive.

Everything can be going fine for a long time, and then cloud costs kill your business.<p>This happened to Murfie a couple of years ago, and that&#x27;s why I had to step in to try to fix things. I&#x27;m still trying, and there are still challenges, but I won&#x27;t allow landlords and cloud costs to disrupt things again.

Think about how many big companies struggle with his. Most don’t have one person who can think through the cost of the cloud, as well as the activities to manage the costs. Many even say “Let engineers be engineers, and business people own the costs.” And all of a sudden you get a ton of surprises…

If Microsoft doesn&#x27;t show the decency to forgive that bill, i&#x27;d be happy to chip in!

Cloud providers should really start protecting customers from these spikes. Alerts are not enough, there should also be hard caps (stop serving) and soft caps (serve at reduced speed&#x2F;capacity) based on configured max budgets.

If you are not a VC backed corporation you must be insane to run anything on a &quot;cloud&quot;. Why not rent a dedicated server from OVH or others where you can actually control costs and pay 10-100 times less?

Seems at least a little unethical that cloud companies do pay as you go up to infinity, instead of some model where you transfer money in and if you use it all up your service gets cut.

As I spent a few hours to successfully get cf cache b2 files. I&#x27;m curious about the part of support Cloudflare requests due to caching issues.<p>It&#x27;s time for cf to work a bit on its UX

&gt;This was about AU$350 a day for a month. It really hurt, and it shouldn&#x27;t have happened. I should have picked up on it earlier and had safeguards in place to ensure it didn&#x27;t happen. It&#x27;s on me.<p>Uh no - it&#x27;s on cloudflare and azure. Why don&#x27;t they have a global setting that says Max Charges Per Month: $X and it just shuts down when it hits that number? This is why I don&#x27;t really like using big cloud services like this.

This prompted me to go and check my custom static site generator (which renders my blog onto an Azure storage account exposed via HTTP and Cloudflare).<p>Turns out I wasn&#x27;t setting x-ms-cache-control when writing all the blobs, so that&#x27;s a win right there.<p>(interestingly, it appears that rclone, which I was in the process of moving to, doesn&#x27;t do that, so I might have to keep my custom Azure storage library around)

Shouldn&#x27;t lookups be where cdb shines? Hold my beer:<p><pre><code> $ shard=&quot;$(echo &quot;${sha1}&quot; | cut -c 1)&quot; $ cdb -q pwned-passwords-v8-sha1-${shard}.cdb &quot;${sha1}&quot; </code></pre> But as a cloud evangelist at Microsoft, you may sing the corporate IT gospel anyway. ¹<a href="https:&#x2F;&#x2F;mro.name&#x2F;agakdfa" rel="nofollow">https:&#x2F;&#x2F;mro.name&#x2F;agakdfa</a>

Well ... it&#x27;s not like it was the first time this happened to a software developer.<p>He should have known better that there is a risk, that you don&#x27;t know some detail that costs you a lot of money.<p>Cloud Bandwidth is soooooooooo expensive. If there is a risk that you have to pay this, please us a provider like Hetzner with fixed costs. If you like your serverless things, just host the big files at Hetzner.

&gt; I always knew bandwidth on Azure was expensive and I should have been monitoring it better<p>It&#x27;s suspicious that cloud providers STILL don&#x27;t have any sort of &quot;circuit breaker&quot; infrastructure for this sort of thing - yes, you can set up alerts, but you can&#x27;t say, &quot;shut the whole thing down <i>before</i> the costs go above a certain threshold&quot;.

I guess all Microsoft PR and Marketing departments are now on the phone trying to get this guy a refund and take down this post :)

&gt; I, uh, have a bill I need to pay<p>Kind of sad that service we are accustomed to using, various software integrates it (whether using HIBP API or downloaded pwned passwords archive) - is on a shoulder of single guy that now has to pay for his mistake.<p>Great that Cloudflare helps him with the service, otherwise who knows if we had access to HIBP at this scale?

Hope it is okay and not too much off-topic. I just donated. He deserves it for this service!<p>Fact is that stuff like this can happen. Consider how many variables are in play to determine the final cost of a cloud service it is very much a double-edged sword. Sometimes you cut yourself unintentionally.<p>So now we all learn from this, I suggest we help him out.

Looking forward for the followup post in early 2033 when he forgets to extend the cost alert expiration.

This is a big trap to fall in to. I dont understand why network trafficking is so expencive also in AWS. I once had a 2k monthly bill purely from networking because i accidentally routed a lot of requests through a NAT. That hurt haha. Now i stay away from those things :D

&gt; But these would always cache at the Cloudflare edge node, that&#x27;s why I could provide the service for free, and I&#x27;d done a bunch of work with the folks there to make sure the bandwidth from the origin service was negligible.<p>If you&#x27;re not Troy Hunt or another celebrity with special access to Cloudflare -- I don&#x27;t think you really have access to Cloudflare to do a lot of work with you to ensure that your data gets cached and your egress is minimal, for large files on a very cheap cloudflare plan. (Based on the costs reported by Hunt as catastrophic, I don&#x27;t think he&#x27;s paying cloudflare for a large enterprise plan)<p>(Also, it&#x27;s unclear if caching large data like this is even within the ToS of Cloudflare?)<p>I don&#x27;t think Cloudflare promises to cache any particular URLs for any particular amounts of time (except no <i>greater</i> than cache headers etc; but they don&#x27;t promise never to evict from cache sooner; they evict LRU according to their own policies). Cloudflare&#x27;s marketed purposes include globally distributed performance, and security. I don&#x27;t think they include &quot;saving egress charges by long-term caching your data&quot;.<p>I have a much smaller project, but egress charges for data are an increasingly large part of my budget. I&#x27;ve been trying to figure out what if anything can be done about it. I wish I had a guaranteed way to get ultra-long-cache promise-to-be-within-ToS for very large data files from Cloudflare for a affordable fixed-rate price. (Maybe I do? But just haven&#x27;t reassured myself of it yet?)<p>&gt; In desperation, I reached out to a friend at Cloudflare… I recalled a discussion years earlier where Cloudflare had upped the cacheable size… Since then, Cloudflare upped that 15GB limit…<p>Since I&#x27;m looking for solutions for this same problem (delivering lots of data at very cheap prices), I am finding myself a bit annoyed that Hunt is talking about how he solved it, using tools&#x2F;price-levels not available to most of us who don&#x27;t have his level of access due to position.<p>Interestingly, MSN&#x2F;Azure is part of the &quot;Bandwidth Alliance&quot; with cloudflare, which initially one thinks means there are no egress charges when delivering to cloudflare. (That is what it means for some other alliance members like backblaze). But that&#x27;s clearly not the case or this story wouldn&#x27;t happen, right? Turns out Azure gives you a fairly small egress discount when delivering to cloudflare, and only if you set things up in a non-standard way.

First thing i do is set an alert when costs go over 10$ for any new project. Highly recommend

Clouds are good for quick start and fast grow. But after this phase, you should think about &quot;classic&quot; hosting solutions (multiserver, load balancer, etc.), they could be much cheaper<i>.<p></i> as long as your human admin costs are lower then cloud services

It&#x27;s unconscionable that MS doesn&#x27;t have warning notifications in place BY DEFAULT, so when you start incurring charges <i>e.g.</i> 10x of normal, you get notified immediately. One shouldn&#x27;t have to set these up manually ever.

It seems like everyone is blaming azure when this was an issue with CloudFlare…<p>I get that everyone has an obsession with dirt cheap providers instead of cloud solutions like aws&#x2F;azure. But that doesn’t mean it’s better. Everything has pros and cons.

I&#x27;m sure <i>some</i> cloud providers have it, but they all should have a global, &quot;If my account hits $XXX shut it all down immediately and email me&quot; flag. And yes, that&#x27;s kind of what he did here, I get that.

I wonder if people will start to make shell companies to just go brankrupt when this happens and start afresh with another company. The cloud vendor doesn&#x27;t look too closely ehat you are running right? So this could work.

Most of the clouds have functionalities to manage this. In AWS for example you can create an alarm with AWS Budget to monitor costs by tools&#x2F;service&#x2F;etc. Using a complex cloud without using this is not good practice.

It is good thing to know that this could happen to anyone. I guess that setting limits and alters should be one of the first things that one should do.<p>What would happen if a credit card limit was exceeded, a site would just stop working?

Yeah the problem with Cloud vendors is that if they make a mistake, it will usually disadvantage the customer...not them. I&#x27;m a little biased as I don&#x27;t completely buy into the whole Cloud paradigm.

Cloud seems like a pet tiger - really cool and fun, until it turns on you.

Enjoyed the article.<p>But still, couldn&#x27;t help to get the following lasting impression after reading it: these days being able to click around the UIs of the cloud providers should be a billable skill by itself.

This guy needs to clean up his bio. There seems to be a lot of confusion on whether or not he works for Microsoft when it appears that he is a uhh... reverse pay midlevel manager inter?

These things really should have a AI like alert that is basically “cost is departing dramatically from historical pattern” without the need to set thresholds and the like

Are there cloud services that allow to easily put a maximum budget, to make sure you have no surprise costs like that?

It would be good if he contacts Microsoft about this. Sometimes they will give credits for situations such as this.

Happily donated to Troy. He&#x27;s done more than most to help everyday folks weather these data breaches.

I don&#x27;t understand it. Does a cloudflare edge server sit inside Azure?

close account, cancel card and move on with life before they charge you.

If anything, this shows the insane scalability of the cloud

outbound transfer cost is one of the most expensive things in cloud computing, it&#x27;s much better when you can pay for allocated bandwith.

Just avoid cloud and choose dedicated infrastructure

Didn&#x27;t Troy sell HIBP to Verizon?

Donated! Hope it helps

TL;DR: I got a big bill from my cloud provider, so I used more cloud provider features, to make sure I know before I get the bill; isn&#x27;t my cloud provider great?

Can somebody explain to me why I wouldn&#x27;t just rent a 40 EUR dedicated server from Hetzner with unlimited traffic and gigabit uplink? His 600GB&#x2F;day is way less than what you get over a gigabit link within a day. Sure, sudden bursts would perhaps &quot;throttle&quot; at a gigabit, but according to his article that was only the cloudflare proxy anyhow, so no pain in having that take a few seconds longer.<p>As far as I am concerned, I just don&#x27;t understand why people use cloud services.