Since this one is the bane of my existence at the moment.
Yeah, this statement is technically correct, it is technically free, but you still pay for the VMs that are part of the cluster, and their disks. (Unless they can qualify for some other "free" metric.) But we've been using this, and we've had lots of spans of time where the managed API server will just time out requests. There's no guarantee on the free version; there's an SLO, but our metrics indicate that it isn't being met. (And our support requests were met with "upgrade to a paid SLA"… which we are, b/c we're a business, but if you're looking at this list, that presumably isn't you.)
Despite being very early adopters, we've only had one issue and when we pointed the finger at them...the DO k8s team immediately diagnosed the issue, and it was our own fault. Woops.
Some time ago i compared them against the competition and they fared very well even against GKE:
In AKS, the API server & its backing storage, as well as most of the core controllers are "managed", in that they run somewhere not in your cluster. But then, you hit the issues where they must consume resources, and someone at the end of the day must pay for those.
I'd almost rather they run in the cluster, since those are resources I can introspect & debug.
(In a normal "vanilla" cluster, not a cloud-based one, normally these would all run inside the cluster, unless you configured them differently. E.g., kubeadm builds them into the cluster.)
In a default setup vanilla Kubernetes, you have master nodes which host the master services ( etcd, kube-apiserver, etc. etc.), and worker nodes on the side.
With managed k8s, in most cases, the "control plane" ( what normally runs on the master nodes) is run by the cloud host, usually on shared instances ( Scaleway did a talk and it runs on dedicated Kubernetes clusters, in a namespace per client).
Indeed, then you're subject to limitations ( e.g. the kube-apiserver for you is limited to 1000 MB of RAM and if you need more because you do X, you're screwed), but that's always the case - if you run your own master nodes, they have limitations as well ( based on instance/server size), just usually higher. There's the added risk that another customer might sature an unlimited resources and impact your control plane.
In any case, I haven't hit any limitations with Scaleway's Kubernetes so far and haven't seen anyone talking about having hit one, while this isn't the first time i've heard about Digital Ocean. I don't know if it's due to size or because Scaleway have higher limits.
Now that info aside, they do support cluster autoscaler.
Or maybe it's more appropriate to say that Cluster Autoscaler supports them? This tool (CA) does scaling in both directions and is highly configurable. At least, it is for other cloud vendors so I'd guess it's going to be so here too.
In regular setups, you pay for the resources, then you pay someone to take care of those resources. In the Azure's case, and AWS EKS and maybe google offering too, you don't have to pay that engineer or team of engineers.
P.s. your first GKE cluster is free
Wikipedia does that through bot, adding an "accessed <date>" to external links. I supposed you could do the same here. You could even use Actions to generate a list of entries in need of checking (e.g. haven't been checked in 6 months).
Wikipedia sentence: In 1735 a volcano on the island of Foobar erupted.
Original article content: In the history of volcanic eruptions there have been many different notable eruptions. One occurred in the year 1601 on the island of Foobar. Another occurred 134 years later on the same island.
Does the citation match: Yes.
Wikipedia sentence: Bar Foobar was a prolific American poet.
Original article content: In recent years, the number of athletes who have been added to the American Sports Hall of Fame has increased dramatically. In fact, the Sports Hall of Fame just added its 1000th new member, Bar Foobar, on June 15, 2014.
Does the citation match: No.
Wikipedia sentence: The leader of the Foolandian revolution was Bob Jones Sr.
Original article content: The Foolandian revolution remains one of the hallmark events of the 20th century. It was spearheaded by Bob Jones Sr. who today has a Foolandian holiday named after him.
Does the citation match:
Tried it and it didn’t work? PR. Found a new one and it’s not on already? PR.
Also, it would be a really nice bonus if “date added” was generated from the git commit that added the resource.
Specially if you auto deploy it too
The people who live on GitHub are unfortunately unable to recognize any distinction.
Maybe it's different for services, but lists of software often list projects that aren't maintained.
I've never found anything useful from any of these lists. What I do find useful is people on HN talking about what software/services they use.
I’ve wanted to setup a personal account on Google cloud, azure etc and take advantage of free stuff but I am terrified something will happen and I’ll mess something up or someone will mess it up for me and I’ll be out hundreds on my credit card. Any suggestions here?
The real answer is that if you have an understanding of networks and systems, then the cloud is just a data center with different considerations. Your company should be giving you resources like sandbox accounts that you can play in. This is how you'll really learn.
When the pandemic hit, flights were cancelled, and refund requests were unanswered here in Canada, many were afraid of requesting a charge back through their credit card company. In a country with only two major airline companies and a size that is too vast to travel by land all the time, getting banned from one, if they choose to do it, would make one's life so much more difficult. Tech is not far from that in terms of oligopoly power.
Was there ever a precedent of an airline ban over a chargeback where a refund was warranted?
Not to airlines, which have your full legal name, date of birth, and passport number on file.
Not to big tech either, which are in practice giant surveillance companies and have mastered the art of tracking everyone and anyone on- and offline.
> Was there ever a precedent of an airline ban over a chargeback where a refund was warranted?
I don't know. But many people would not have the time to research this to their satisfaction, nor would want to risk it and find out the answer the hard way.
I highly recommend something like this if you want to learn the cloud and do not have an employer to sponsor an account for you.
All these suggestions of using a throw-away CC number to avoid paying the bill are just unethical, IMO. No need to resort to those measures.
There have been horror stories posted here before.
Not affiliated, I'm in Aus and I use my temp card for anything I don't want to be ongoing. Doubles as a breach safeguard (my credit card involved in a DB dump doesn't matter if it was a one-off number).
There's other good reasons to use revolut but will save it :)
His entire account got locked out until he was finally about to get ahold of someone in billing to reverse it.
It was right around the time that this article: https://smartprivacy.io/learn/i-got-banned-from-google-ads-f... was making the rounds.
Since then I'm a happy free Heroku user, but also tried Azure, Google, AWS. Azure had the worst UI and features, but the others similar bad. Google also cannot be trusted for such a freebie, because of their random PM policies, Amazon forgot why. Both high risk to pay unexpectedly.
(I've had success with this in the past but apparently results can vary - some services may block a prepaid card)
Don't do it, Amazon's whole shtick is having such opaque and illogical and convoluted pricing that you just throw your hands up and let it go.
Any framework would be interesting for learning purposes. I have some older MVC .NET sites and I poked around with Azure free Windows Server VMs, but the performance is so painfully bad; if you do anything, you run out of memory. But I'm gaining proficiency in NodeJS and Python, and hoping to learn how to take advantage of free tiers for practice in development and testing. I imagine it can all be self-taught/learned, but if there exists documentation to go through the motions in the learning process, I think it would help a lot of people.
Heroku is cool, too, though.
But that's orthogonal to the point I was making, and the comment I was replying to. If someone likes Heroku because of it's pricing clarity and its ease of setup... well, the pricing simplicity of an EC2 instance is just as good, and the ease of setup is even better.
More broadly, if you're evaluating them against each other....
Heroku isn't the easiest to set up, nor is it the simplest. It's not the easiest because you have to learn a bunch of Heroku-specific things, that you'll just have to reinvent/relearn/etc if you ever want to switch providers. To call it "vendor lock-in" might be a bit of a stretch, but it has aspects of that. With EC2, it's just server management, same as it ever was, same as it'd be on some other cloud.
(Note that many OTHER AWS services are far worse than Heroku in this respect. And/or better, depending on your goals.)
Heroku aims to be a sweet spot. A few hours (or days, maybe) invested in learning to use it properly, and you get a pretty reasonable bundle of tools and support and so on. But if you want to dive deep, Heroku's probably a waste of time. And and the other extreme, if you don't want to learn anything about an individual provider's proprietary crap, then Heroku also will not work out that well for you. Saying "this is our sweet spot" is a respectable position, even if you don't like the sweet spot.
(Personally, I'm still boycotting Heroku over their years of misrepresenting their "dynamic mesh routing", which was always fiction and which they lied about for years, and charged customers for something that they knew full well they weren't delivering. Don't get into bed with sociopaths, even if you like their sweet spot.)
It’s an understatement, not a stretch.
Heroku has always seemed really interesting/cool/clever to me. The tradeoff has never been worth the investment when I’ve considered using them, personally or professionally. Their operating model is far too different than every other cloud provider’s.
By the end of my 30 days, I found the bare minimum performing low volume .NET + SQL on Azure was going to be $30+/month even for just a prototype/development site, and I can pay less elsewhere for higher performance (though perhaps less control). And when I say "performing", I just mean the VM not being too slow to actually use as a UI, and throwing low memory errors constantly. The free tiers for Windows Server are, in my opinion, unusable at all.
You just have to bite the bullet and learn a few linux admin tasks like setting up nginx, supervisor, file permissions.
By .NET + SQL, I assume you mean you were using an App Service? They come with a lot of convenience, but I definitely agree that they are too expensive, especially when considering the level of compute you get. I raised this with someone from Microsoft before, and they said they were confident the value offered was worth it.
There is a free tier for App Service though. It's a shared compute model - I tried it a while back, and the performance was totally fine for light usage.
I wouldn't try to run Windows Server on a B1S, as they only have 1GB of RAM :) You might just about get away with it for Server Nano, but I've never actually used it. Best stick with Linux or BSD for these tiny VMs.
Edit: also, if you find that you need other cloud offerings from AWS or Azure, you can combine it pretty easily. I did image upload to S3, for example, from a Heroku app.
What I'm trying to say is that you do get what you pay for (or don't).
Over the last several months we've migrated most things to Go (which was always the ultimate goal), and introduced a plug-in based approach with it. Recently we enabled a feature to allow you to enable or disable plug-ins used by the instance agent dynamically.
I haven't had time / opportunity to look in to Microk8s or k3s. Would it be straightforward for me to replicate what you were seeing? Just follow a simple online tutorial?
Note: We built/build OCI with security in mind, and encouraging a more secure-by-default approach. All of the standard platform images come with the firewall enabled and locked down, and the default network security groups are locked down too. One example brought up in the earlier days of the platform build out were all those times you see people's MongoDB accidentally exposed. Obviously this is a total guess as to what you ran in to, it's definitely something that trips up customers.
It'd be really great to show developers they have a place they can make toy clusters without having to worry about gotcha charges.
Would be interesting to see a comparison of Oracle cloud to AWS/Azure/GCP, from the perspective of a developer or architect. If of course Oracle allows such comparisons in their terms and conditions...
Gold, Jerry. Gold!
Using standard / paid instances should be no more difficult or hard than using free-tier ones, other than the obvious need to have converted to a non-free account and, I believe, sorted out payment methods. They launch and run the same way, just when you have a paid account you can launch on larger instances. We've got enough things we want to do with the cloud, without having to resort to building lots of separate experiences and code paths for free tier accounts.
Disclaimer: opinion is entirely my own, may not reflect the opinions of my employers, etc.
The SDKs are open source and can be installed fairly easily, e.g. the python sdk is up in pypi, as is the oci cli.
Vendors are welcome to sign up and start publishing images via the marketplace on our platform. That's an option available today, without specific need for OCI to engage. SuSE are in there, for example.
It's an awesome deal for basically 0$ bills every month (we pay for a few other things though) and great for us as we are a small startup bootstrapping at the moment. It more than handles our development and early access traffic.
This 0$ setup gets us highly available docker + firestore for absolutely nothing until we hit the freemium threshold. We went over by a fraction of a dollar on a couple of occasions when we did big imports, etc. But we won't start paying until the point where we have recurring revenue and users basically causing increased traffic.
I also like not having to deal with devops for this. No faffing around with doing terraform, setting up CI, writing complex deploy scripts, etc. Frees up a lot of my time. If you have done any of that in the past years, this is shockingly easy in comparison. A simple wizard, couple of clicks and you have your docker container building and deploying straight from github. I was not expecting this to work and but then it did on the first try.
Switching the whole setup to using vms is super easy to do at any point since you can basically just tell it to run a docker container. It's nice to have that option. But then you need multiple vms, a load balancer, etc. and before you know it you are looking at more like 75/month. Add a decent database and it gets worse. High availability, worse still. And you need to do the devops to connect all these bits of infrastructure. You will blow more on staffing cost than hosting. Even a single day spend by a senior that can actually do this would cost you like 1000$. Lets be fair, it's never just a day.
IMHO this is what most deployments should be like: bog standard docker thing, go run it and take care of all the things that aren't optional like giving me some dashboards, alerting, etc. and just pay for what you actually use.
Does Cloud Run's free tier 'sleep' containers when they don't get regular traffic?
Several years back Heroku nerfed their free tier -- they would proactively place low traffic apps into 'sleep' mode so the next request would have a noticeable several second long delay.
I ask because our SpringBoot monolith takes 30s to startup, and has bursty traffic, so I was of the expectation that Google Cloud would probably 'sleep' low traffic containers like this also. Would be nice if that were not the case.
I actually only set the minimum instances to 1 a few days ago as we at this point are preparing to onboard some customers.
This allows you to get around the credit card wall, without fear of them actually being able to charge the card.
I also use this trick for any annual membership type things. I will put in a virtual card number to pay the annual fee, then after they charge the fee I will deactivate that virtual card (but not delete it). Now if a year from now they charge the card without warning me, it will decline. They usually send you an email saying it declined, at which point you can either leave that card deactivated and you now easily cancelled the service, or you want to keep the service you can go and activate that card back up and the next time they try to charge the card again (usually 48 hours later or so) it will go through.
If anyone knows of other ways to create virtual cards, please let me know. I am currently using Capital One's virtual cards because 1) it is free if you have a Capital One credit card, 2) it allows unlimited virtual cards, 3) their browser extension is actually really slick at auto-generating cards and filling in the card numbers for you. I have looked into Privacy.com which actually offers a lot more granular control over your virtual cards. But you have to pay monthly for the service and get a limited number of cards, so I stick with Capital One. I have a Capital One card that I don't use for anything other than online subscriptions and virtual cards. But I would love to know if there are other options out there that I could consider.
This is definitely a limiter for me when signing up for these free sites: it usually is couched with language along the lines of you don’t want your business to be impacted by us turning off your service, but I am not joining as a business, I am joining as a “I am a developer investigating if this is worth it for my business”
I would prefer plans along the lines of “pay us $10 to activate, if you go over traffic we cut you off but guarantee no other charges, if you leave we refund your $10 no questions asked” or “no charge upfront, if you go over the free tier we charge up to $10 once only and then cut you off”
privacy.com (https://privacy.com, or If you want to use a referral, privacy.com/join/NPNDJ ) is also free and works across any bank/debit card, although just recently they started monetizing by limiting free accounts to creating 12 virtual cards a month, so if you need more it's $10/mo.
> This allows you to get around the credit card wall, without fear of them actually being able to charge the card.
Note that services can see if a card is virtual and might block them from signing up for free trials. Digital Ocean, for example, does this to prevent referral fraud. Netflix also used to block virtual cards from signing up for the free trial when they had one.
It's not clear to me what you mean here: do you suggest SaaS businesses should reject those customers who prefer paying with prepaid cards? Hello, this is me! When I see a prepayment option or even Paypal, I gladly pay, but when you want my credit card info, I wave you good bye!
Not the parent, but: yes, absolutely. Prepaid cards are for making one-time payments. They are not suitable for setting up ongoing subscriptions, and it is entirely appropriate for a merchant to reject them in this context.
Why? Because from the point of the view of the customer, they are the best fit for that purpose.
From the point of view of a legitimate customer who intends to pay for a service, a prepaid card is a poor fit for an ongoing subscription, as it will stop working unpredictably when its funds are exhausted (causing it to start rejecting all charges).
Optimizing for customers who intend to ditch a service without paying for it is not a goal.
A simple example of one of the services I use, Mailerlite. When the funds are depleted, I receive an email from them they were unable to charge me this month and they downgraded me to their free plan, and they will try again in two days. I then charge my card, they charge me and everything is fine. You don't want people like me - fine, you will have less customers. And I'm a very loyal customer, cases like failed payment happen maybe once 2-3 years.
To extend the plea, if anyone knows how to do this in Canada, I’d love to know. As far as I know the services that exist are USA only.
However, one service that is doing this is Revolut which has a private beta for Canadian users and will provide virtual cards[1,2].
Nothing that mpv with youtube-dl couldn't solve, but still.
These people have no shame. Frankly, I wonder how far they will go.
But let's also not forget that Google is (was?) parsing your banking info from your emails with transactions and purchases, and there wasn't even any option to turn it off.
Can anyone explain why would you want it as a service?
It is intended as a "next step after hello world" example for an unoriginally named low-code integrations platform, rather than being someone anyone thought someone might need it its own right.
Brian Called Brian: You don't need to follow me! You don't need to follow anybody! You got to think for yourselves! You're all individuals! You're all different!
Man: I'm not...
A website asks for your HN username, validated the age, gives you a random string to add to your HN about page.
Once done, your accounts are now linked!
Perhaps that is why I’m the entrepreneur ;)
But they make me feel like a boomer. Why gets their info through plain text?!?
My app has 300 features. Beginners get to use 8 of them. Once ten people think you are not a wretched toad, you get to use 20.
Once you've solved a problem for 200 people, you can edit other people's posts.
In fact doesn't HN not let you downvote until you get 500 upvotes? It's been a while.
I am a Wikipedia auto-confirmed user, so that lets me make articles without needing to go through the process, so that's nice.
The problem with all of these sites is that they don't also police the opposite side: the obsessive guy who spends all his time on the site and attaches all of his identity to it so he is just a negative person despite having lots of 'karma'-equivalent.
It's like the value-to-community vs time-spent curve is like
| - -
| - -
| - -
| -- -
| -- -
I'm medically handicapped and "have no life" which is why I spend so much time online. When I first joined HN (July 2009), I figured everyone on the leader board here was basically some loser with no life.
Not a big slam because I figured they were fairly intelligent losers. But I did figure "You people are spending so much time here for the same reason I am: You desperately need to get a life. (Only you seem to have less excuse than I have for your failure to get a life.)"
That proved to not be an accurate assessment of the situation.
Your point is not irrelevant, but I will suggest that karma count on HN is not a good measure of anything at all other than time spent actively posting on HN -- which, in some cases, is cumulative over many years and in some cases it is not.
Some people here certainly are "the town drunk." And some people with high karma are people who have been here a long time and contributed a lot of good stuff while setting their ego aside and so on.
My intuition is that this is less of an issue on HN than on Wikipedia.
Conveniently we have a natural experiment from a year or two ago we can refer to.
Stellar Lumens offered some free cryptocurrency to existing users, via Keybase accounts that leveraged external auth-equivalence (such as HN). It backfired spectacularly for all involved. 
 I think this is it - keybase.io's original blog post (referenced at  ) is 404'ing.
Edit: Ah, I see, people tried to find dormant HN and GitHub accounts. https://www.coindesk.com/stellar-tried-to-give-away-2b-xlm-t...
This is kind of gatekeepy coming from an account that's not 10 years old yet. :-P
Meh. I made up a threshold I thought might keep abuse of a free service down for a testbed. I picked 500 since I thought that was the HN threshold for downvote ability. Sure seems to have stirred up the crowd though :)
Their section on Data Visualization on Maps lacks a link to Stamen Design free tiles, available under a creative commons license.
I'm wondering if people here think it's worth submitting a pull request or if that sounds like it is outside of the scope of what they are trying to curate for some reason.
While the tiles are free (CC) I would say a list of freely licenced downloadable resources/assets for devs would be much bigger/broader in scope.
For me, Stamen Maps have been a big deal for stuff I do. It took me forever to find something of this caliber that's suitable for my aims, but I'm just doing maps for various projects and not some kind of GIS thing.
To distill this down for the purpose of this specific topic (map tiles):
- the map tiles being provided under CC is a copyright / right-of-use / legal intellectual property thing. Nothing to do with any provision of material service by Stamen
What Stamen provide that's relevant here is a service, that's somewhere vaguely in between SaaS and PaaS.
- SaaS would be if Stamen provided some kind of maps website, with accounts, etc. for working with data & displaying maps. They don't really do that fully.
- PaaS is a bit closer to what Stamen do: they run a server and host their own map tiles. You can point your website at their server, and they pay the hosting cost of the traffic you send there. (PaaS as a buzzword technically involves a bit more than this, but this would be a subset of it).
- IaaS is an unrelated thing here: this would be like if they offered servers for you to go and host your own map tiles on, along with a load of tools for managing those servers.
TL;DR: The CC-licencing of tiles is about releasing "artwork"/IP? (is there a better term for map styles??) that they've already created. Tile-hosting is about paying for internet traffic of the browsers hitting the server where those tiles live.
While I've known of some of the free tier things on AWS, and Google Cloud, I never made use of them long term.
I use them to host a bunch of misc. scripts/programs I've written, a discord bot I wrote for a server I'm in with some friends, my calibre ebook library is also backed up there and the calibre content server is running so I can access it over the internet should I want something to read and be away from my computer.
Mostly good experience, if not the most user friendly. The data transfer allowance on their free tier is fantastic, which is definitely the exception from most free offerings I've seen.
I also use Cloudflare for my static website on S3.
> (note: You must pay if you use .casa, .cf, .click, .email, .fit, .ga, .gdn, .gq, .loan, .london, .men, .ml, .pl, .rest, .ru, .tk, .top, .work TLDs due to spam)
Makes me want to search my spam folder for .gq out of curiosity alone.
I'm not bagging the whole thing by any means, it looks like a really useful list but some things are a lot easier to understand what you're getting for free than others.
They couldn't make it more user-hostile if they tried. (Maybe being hostile is the point? TBH I don't know the real monetization strategy behind Youtube, maybe hostility pays here.)
I highly recommend it.
A lot of times we aren't event aware that there exists a tool which can solve your need. And Google is of no help here.
Edit: from Sweden if it matters
The site works fine.
could just register that and make the subdomain free.
I'm saying this as someone who knows way too much about Oracle databases ;-)
Choose a "puppy" you can love. Get it from someone you trust and get one that's right for your situation.