I've done very little with AWS for two reasons - haven't had any projects/work where they used it, and personally...
Several years ago I signed up for a free tier account with AWS to play with . But I also didn't know what I was going to do with it - I didn't create any resources. Got distracted, a few weeks went by, they sent me an invoice! It wasn't for much, but it was not $0. I was very confused (and never figured out why I was charged). Fortunately I was able to reach customer service and get the costs dropped, and I promptly closed my account so that I didn't accidentally (through no action of my own) accumulate further charges!
I'd like to see "get charged more than expected" or "think you were on the free tier but somehow you weren't" on this purity test.
EDIT: OK either "free tier" was added just now or I fat-fingered "free" when I searched for it earlier :)
The blank cheque billing is one of the reasons I don't use AWS for any personal project either i.e. they do not allow you to set a hard account cost limit. The billing approach makes AWS scary to learn on your own time as you don't know if you're going to get a huge charge from clicking some button or other you didn't understand properly. I deal with AWS reps for business and they don't seem to perceive it as a problem when I have mentioned it (or don't want to do anything about it).
They don’t perceive it as a problem because it’s not a problem. You getting charged 25$ by accident is not even worth thinking about considering adding a meeting to discuss potentially adding a ticket to a jira board.
If you want fixed charges use AWS amplify or another service built on AWS primitives. But the whole reason for using AWS is flexibility and “pay for what you use”.
If you don’t understand what a button does then don’t press it until you understand it?
Flexibility and pay-as-you-go has absolutely nothing whatsoever to do with the capability to set an account hard cost limit. If AWS wanted to add such a feature they could do. They obviously don't for their own commercial reasons. It is very easy on AWS to make mistakes and it's off putting to people like me who do not have unlimited budgets as clearly you do.
How would this actually work? Actually think a bit about how hard it would be to add a global, synchronised and immediate view of an accounts cost and check it on every single API call.
Take S3 for example: you set a 1$ limit then concurrently upload 10 1$ per month files. What happens? Should 9 of those uploads fail? How would you ever synchronise that? Or perhaps they complete, but should the bucket delete the contents after the ”storage cost per hour” limit is reached?
Perhaps you could reconcile the account cost usage every hour, and when reached the account should be locked somehow and no new writes or reads can be done on anything. Which is basically breaking the account and everything that runs on it. Most people don’t want that.
How would it even work with bandwidth? You’re serving a file from S3 or EC2 with your 10$ account limit. What happens after you use 10$ bandwidth? Should your accounts external internet connection be cut off? How would you even calculate that cost in real time at scale.
Use your brain. AWS are not stupid. It’s a minefield of complexity that would be super expensive to implement and that benefits nobody. The “if you mess up we will refund you up to 2 times” policy works absolutely fine.
You're being aggressive and rude now, so I think this will be my last reply on this subject. As to "benefits nobody", that's clearly wrong. I have clearly stated what the issues are and other people on this thread have raised similar issues with AWS billing. I don't think you'd have to do too much googling to find more people with the same complaints.
What they could do is pause services immediately that can be paused and then give you grace on services where their billing technology has not caught up with real time. They don't necessarily have to actually stop services if they cannot technically do it instantly, just stop charging you at an agreed amount. That's all that's needed.
Sounds like a great way to dodge your expensive AWS bill and get a bunch of resources for free.
“Oh sorry I used 100tb of bandwidth over my 10$ limit, I guess your billing technology has not caught up with real time! Lucky you are so nice and stopped charging me at an agreed amount”.
>Perhaps you could reconcile the account cost usage every hour, and when reached the account should be locked somehow and no new writes or reads can be done on anything.
attacks are frequent, and then you gotta deal with waiting on aws to confirm it was an attack and refund etc. it's easy to avoid attacks with best practices but the fact stands that they are still common in customer accounts and then you have a financial risk on top of the attack risk.
The reason I don't like AWS is because billing is always at least like 24 hours behind. By time you realize you don't goofed, the bill's already probably more than you thought.
That said, I'm vaguely amused at the concept (I got a 23), but they also misspelled my name (Corey) and the name of the newsletter (Last Week in AWS, not "This Week").
I also have some quibbles about a few of the questions, but that's probably no surprise...
Yeah instantly my thought. Also since it’s only in “request access” mode it’s not something that can be ticked off eg not really apart of the list. Also the landing page for it feels like it’s written in the same style.
Definitely not against advertising the product but make it more explicit
If the author(s) of this website are reading: The LinkedIn info I found about you was all Web3-related. If you're actually going after people looking to reduce their AWS costs, you may want to fix that.
Good to know it’s not just me (but seriously, the way they’ve implemented search in paginated results is the most actively hostile UX decision I’ve ever encountered. It’s so bad it almost makes me feel better about the half-baked hacky stuff I’ve done over the years.)
I will offer a very tepid "I've seen worse" for AWS's UI considering all the things it has to support.
That said, the redesign made it worse (those scroll windows that don't fill up the pane... MY GOD HOW DO YOU NOT FIX THAT), there is no sense of continual improvement and addition of useful new features and views, performance is pretty bad.
All I ever hear "you're not supposed to use the UI, you're supposed to use the API".
Don't get me started on the API.
This quiz is missing these three key points:
- has culled 1,000$ annual savings in a AWS cost saving review?
- has culled 10,000$ annual savings in an AWS cost saving review?
- has culled 100,000$ ?
- has culled 1,000,000$?
I ... almost .. got to 1,000,000 once.
- trapped AWS support in an outright lie?
- gotten AWS support to tell you results of hidden internal metrics to debug something?
Yeah I'm always surprised by the downright hate. I use the console for 95% of things and it's alright.
My gripe is that the thing just gives off this BLOATED impression. Safari routinely reloads it because it was "using significant energy." Every time I load a page I just wait until it renders, goes blank, re-renders, goes blank, places everything that was on the left over to the right, and then finally things settle down and I click a button. That all takes about 1.5s though, so not awful.
Ah, read API as CLI (with AWS they're essentially the same).
The biggest problem with GCP's api/cli/sdk/console is they're not the same. There are things you can only do in the console, or have different grammars across CLI/SDK, or have different permission schemes across them. Add to that the intermixing of "IT Admin" and "Cloud Admin" roles due to all of your Google Groups also being IAM groups, and IAM being somehow tied into your G Suite admin area, it's just a confusing mess.
With AWS, the CLI is the API is the SDK. So if you can use one, you can use all of them very easily.
> the way they’ve implemented search in paginated results is the most actively hostile UX decision I’ve ever encountered
Who is the "they" in that sentence? Each AWS product team writes their own console UI, as far as I know, and they're free to do pagination however it makes sense for them.
(Of course my question doesn't really matter because they're all pretty middling.)
I don’t know but the behavior seems consistent across services. When there are multiple pages of results for something (running ECS services, whatever) if you filter the results it doesn’t coalesce the results, you have to just keep clicking through multiple empty pages, sometimes dozens of times, before finding a result that matches the filter, or even know if anything matches.
Most of the underlying APIs don’t have any concept of filters and most do pagination and will only return a maximum of 50 results. Given those limitations, it’s the best you can do on the UI.
Am I the only one that likes the AWS console interface UI? My only complaint is that some features/products don't really have a UI or require you to manually create the internal IAM role before you can use it.
IMO it's a better UI than O365/Azure and I don't have much experience with Gcloud to say much but they seem about equal but still leaning towards AWS as the best UI of the top 3 cloud providers.
OMG, as bad as the AWS console is, Azure is much worse. The convoluted interface is bad enough, but the SPA warts make it even worse. Not supporting basic browser functionality like tabs is awful.
Surely the console is only for experimentation and checking on things? I figured anything proper should be done in a reproducible way (IaaC like Terraform/CDK/CF)
While we're joking about AWS adventures, has anyone here seriously contemplated using S3 keys as a cheap blob store? It looks like they're a generous kilobyte a pop, and prefix queries are supported, so you could hypothetically chunk your data into a [key|index|data_index| scheme and store a bunch of empty files in slightly under 1KB chunks -- ingress is nearly free (API calls, though you can batch them to keep this from being terrible), egress is expensive, and storage is free. Does that work?
This is a terrible idea. S3 charges $0.005 per 1,000 PUT operations. So let's say you want to store 1GB (1e9 bytes) 1,000 bytes at a time, that's 1,000,000 keys.
$0.005 * (1,000,000 PUT operations / 1,000) = $5.00 to write 1GB. The on-going storage cost I believe would be free since the files themselves are empty.
Meanwhile writing a single 1GB file to S3? Costs you $0.023 per month storage, and the single write cost is $0.000005. It'll take 217 months to cost $5.00
They changed the pricing dimension when it was in beta.
They've also raised SNS prices in some markets where regulatory fees increased, boosted pricing for domain registration several times, redid the pricing model for IoT Core when it was in beta, and changed the pricing model for AWS Config in ways that are kinda... unfortunate, for a lot of us.
If you track the extortion that AWS charges for egress bandwidth against both inflation and what bandwidth costs on average then they are always raising that rate.
Having worked at AWS for about half a decade in the past, I can guarantee that if you've thought of a way to abuse^W leverage their services for alternative purposes, someone has thought of it internally, as well. I knew someone back in 2014 who was working on a project to build a data store on tags, since tags were free. (The reality is that the tagging infrastructure would never have supported this kind of large-scale use, though. I remember a customer who was applying multiple unique tags to every single EC2 instance they launched, and by the time they had launched hundreds of thousands of instances, they were seeing very real latency when interacting with tags. I think the underlying performance issue was resolved, but these sorts of gotchas are all over the place.)
> Has woken up in the middle of the night to a billing alert email?
I went through an awful period of on-call at my last place, and seeing "Amazon" come through on the work email shocked me awake thinking it's another CloudWatch alert, only to blearily see it was the regular end-of-month statement.
You get the same feeling when you wake up before your alarm and you think you overslept, but now you can't get back to sleep for the adrenaline in the blood.
My first thought was "surely the hacker purity test predates that" ... and then I did a little bit of searching. The Hacker Purity Test is from '89... http://www.armory.com/tests/hacker.html
> The Rice Purity Test is a survey first administered by Rice University in 1924. The test asks 100 yes-or-no questions to output a numerical score, with 100 being the “most pure” and 0 being completely “impure.” The first test was given only to women in order for Rice to gauge how risqué their female population was. It often resulted in shaming: women with scores below 70 were often viewed as immoral, whereas those with scores above 90 were seen as uptight. The test has since become popular worldwide, despite retaining its most fundamental problem: it reduces test-takers’ complex experiences to a single numerical value.
Correct. I did the same search before I posted it. =)
Side note, I remember finding one of the dirtier purity tests on usenet after I got onto the internet in `91. Just being fresh in college at that point, a few of the terms used were new to me. hahahahah.
Ah that’s freaking great. I remember taking the purity test when I matriculated to Rice (decades ago). I remember having one of the lower scores amongst the good folk, but not obscene. Something like maybe a 50 or so… always good to see how one is progressing on their goals.
We have a unique one: back before the z1d launched, we had a critical system that was very single-threaded and its performance drove revenue in an extremely direct way.
So we were, deep in the shit, fighting fires due to our systems being in a perpetual state of DoS, and we asked if they would rack a machine with a certain overclocked Skylake chip in it. We offered 7 figures to do it. It was worth that much to us to get ~40% more capacity without having to migrate off AWS. They said no :)
FYI, my teenage daughter informed me this is a parody on a Purity Test going around at secondary schools. That purity tests is regarding romantic relationships and all its aspects.
There have been many "purity tests" before and after that one in the exact same format, that one is just perhaps the most well known so the author copied the format.
I'd have thought that it's more likely a parody of the now quite dated Hacker Test (itself inspired by purity tests) that was making the rounds on Usenet in the late-'80 and '90s: http://www.mit.edu/people/nathanw/tests/hacker
Random reflection: Key to AWS costs is to always assume every compute, storage or networking unit incurs cost. Always.
Then got to (a) pricing of the service in question, (b) FAQ of the service, (c) pricing and FAQ for EC2 specifically and scan for every pricing information you can.
If after that you can be 100% sure something does not incur cost, only then you are allowed to remove it from your financial model of what you want to deploy.
I'm disappointed there wasn't a specific entry for "sat on a beach in Hawaii on Christmas vacation having a 3-way email with the CEO of my startup (who is sitting on a beach in Tahiti) and an AWS contract person to finalize the signing of a 3-year multi-million dollar contract". That's what I did.
I remember back in the 90s lots of these purity tests circulated on AOL and elsewhere on the 'net.
I thknk there was at least a "geek purity test" and a "programmer purity test".. and maybe a "UNIX purity test"? Not sure. They were all quite funny though (at least my younger self thought so)
On-prem gal turned AWS because people don't hire you without cloud. It's not as bad as I had assumed. It sure is a lot of bullshit but no different than working on-prem, just different kinds of bullshit.
Like how-fucking-dare RDS use DNS for failover instead of ARP so it's impossible to do instant non-user-visible recovery.
I was surprised it didn't have things about the number of reinvents attended. I only went to the first one and it was enough of a zoo that has only gotten zoo-ier so I have not been interested in going again. It was certainly worthwhile at the time, but I can't imagine how over-loaded they are now.
That's for private link. Direct transfers in the same VPC are still $0.02/GB (in+out). Privatelink has to be used together with an LB, which is not free (hourly+byte processed)
Society doesn't need to change just because you don't like this specific sub-genre of joke. Romantic partnership, and subsequent occasional annoyance with said partner, is an almost universally shared human experience.
a) you are (wrongly) assuming that this one person is the only person who dislikes this type of joke. You could make this same (wrong) argument to discount literally any suggestion or criticism - "you are the only person who feels this way, therefore no".
b) just because something is common or widespread, doesn't mean it should be normalized or treated as "the best that it can be". That's a highly conservative (in the non-US-political sense) and neophobic viewpoint - not what I would expect on Hacker News. Reduction of a S/O to a sex-provider might be common among certain social groups, but that's not the society I dream of.
Nothing in the joke said "only". By that logic, all single sentences are reductionist.
You're intentionally using emotionally laden words.
Also, the difference between a BFF and an significant other generally is the sex.
> you are (wrongly) assuming that this one person is the only person who dislikes this type of joke
Making everyone happy isn't possible and it's generally not even worth trying because some people just dislike everything. If they're overly dramatic about the "harm" it usually means there isn't any, that they just like controlling discussion.
> just because something is common or widespread, doesn't mean it should be normalized or treated as "the best that it can be"
Nobody did that. They said that in a sense of "and yet you're not dead yet."
Generally humor like this is only bad if it unreasonably hurts someone and "all SOs" are not a class who can be hurt by humor. Nobody is being maligned.
Good point. Our intentions were to keep this shareable and appropriate within a work context. We took inspiration from the original rice purity test, which is very NSFW - but missed the mark on this question. Fixed
Being a good person doesn't simply mean "avoiding whatever linguistic pitfalls the boogeyman SJWs have thrown in your way" - it means actually thinking about the impact and implication of your words and actions, even when there isn't a convenient buzzword-criticism that could be applied to them.
Several years ago I signed up for a free tier account with AWS to play with . But I also didn't know what I was going to do with it - I didn't create any resources. Got distracted, a few weeks went by, they sent me an invoice! It wasn't for much, but it was not $0. I was very confused (and never figured out why I was charged). Fortunately I was able to reach customer service and get the costs dropped, and I promptly closed my account so that I didn't accidentally (through no action of my own) accumulate further charges!
I'd like to see "get charged more than expected" or "think you were on the free tier but somehow you weren't" on this purity test.
EDIT: OK either "free tier" was added just now or I fat-fingered "free" when I searched for it earlier :)