Heroku Free Alternatives
github.com
github.com
https://render.com/ - One of the top Heroku alternatives with a free plan to get started.
https://fly.io/ - Run your full stack apps (and databases!) all over the world. No ops required.
https://railway.app/ - Railway is the cloud that takes the complexity out of shipping software.
https://www.cyclic.sh/ - Connect your GitHub repo. We will build, deploy and manage the hosting.
https://qoddi.com/ - Qoddi is a fully managed App Hosting Platform running on a tier 1 network at 10% of the cost of similar solutions.
https://www.deta.sh/ - Build & deploy your ideas on the universe's most developer friendly cloud platform.
https://adaptable.io/ - Just connect your GitHub repository and let Adaptable handle the rest.
https://www.alwaysdata.com/en/ - All your services in one place.
For example, we had a monitor set up in Papertrail that notifies us if it doesn't see our application's heartbeat log entry in a 10-minute period. We had to disable that monitor because Render will regularly have outages related to logging where logs simply don't make it to the log stream. Their status page is all green and I've unfortunately had to reach out to support on more than one Friday night.
Deploys take a little too long, around 5 minutes on the paid Pro Plus tier.
And, minor gripe, but the web UI feels like an SPA, and I think it's built to be one, but the frontend state isn't stable and will often show the wrong information. From time to time, you'll need to manually refresh the page to get the correct information.
The security issue that happened with Heroku back in April and the subsequent flubbing of communication around the issue made it so we had no choice but to migrate. But the platform itself was rock solid in my experience. I don't get that same feeling from Render.
I'm holding out hope that it improves in the coming months!
EDIT: I double checked the deploy times because of the comment below, and it's actually closer to 5 minutes, not 10, so I've updated my comment accordingly. 5 minutes is much more reasonable.
For example, we've worked hard to get logging to a stable state over the last couple of weeks, and I hope you've noticed the improvements already.
Regarding deploys, I'm surprised they're taking this long on paid plans. I'd love to follow up if you could share details over email (address in profile). We have internal SLOs around build times and we've made (and observed) improvements recently, so I'm wondering if this is specific to how your build is being cached.
I double checked the deploy times and it's closer to 5 minutes, so I updated my comment accordingly.
I'm rooting for you!
However, they delete your database after 90 days if you do not have a paid plan. From https://render.com/docs/free#free-postgresql-databases
"Render’s free database plan allows you to run a PostgreSQL database that automatically expires 90 days after creation. Free databases are suspended after 90 days (unless they are upgraded to a paid plan), and are no longer reachable at this point."
The paid plans do start at $7, and you can share a database across many apps, though.
And if you have static site or a site without a database, it looks like it is forever free.
It’s still behind the waitlist but it’s coming off in a few weeks. Also if you onboard via hasura.io there is no gate.
Could you bump the email in my inbox? Even if you hate my guts and have no interest in the company, I‘d like to know cause I don’t recall this at all :(
We do a 30 minute screen with me (to protect our engineers time) and share the problem ahead of time (to not surprise the interviewee)
They do seem to be pushing some sort of creative-focus, maybe they intend to just keep it really small user base, invite only, just people who actually make things vs a platform just to run your SaaS on?
https://northflank.com/ - The comprehensive developer platform to build and scale microservices, jobs and managed databases with a powerful UI, API & CLI.
I will say I’m not sure how the reporting is supposed to work. Fly thinks I’ve been running a failed process for the last couple days but it sure is working!
- Latency of the oregon region is very high. A simple health endpoint is around 500ms from north america and europe, 1.3s from Asia/Australia (we have been monitoring it for many months with better uptime)
- Services tend to die at least once a week, so you do need at least two instances all the time
- Logs are often delayed for a long time, though it has been more severe recently.
- Build time is just horrible, our rust service takes around 40m to deploy. I get that it is free, but at least give me a docker registry so I can build it elsewhere and push the image
- No option to pay for support 24/7, so if there is an outage during the US night you are on your own
I'm surprised to hear about the service dying once a week: in my experience this is typically an application issue. Happy to help debug if you'd like to shoot me an email.
I wrote about the experience here: https://onlineornot.com/on-moving-million-uptime-checks-onto...
I am not saying that it fails to provide clarity to someone who is willing to pay for it, but the rule of thumb of pricing pages offering a "free" product is to explicitly mention this ⇒ "$0". Am I being picky? Yes. Why? Because it is a Heroku alternative, and Heroku did mention free in an obvious way in describing their pricing plans.
When you mention overage charges immediately after the free tier, it kinda feels a little bit weird. In the pricing plans section, they don't mention a free plan.
What’s important is that pricing isn’t the same and the free plan is really a way for you to try Fly before deciding to dial up the resources. You can use it for really small apps, but trying to squeeze a Rails app into 256Mb RAM on 3 servers is tough.
I moved few of my Heroku apps over to Fly and pay about half much for about 4x what I was getting on Heroku. When v2 of the Fly app platform goes out, I expect that price to go down even more since apps can sleep when they’re not serving up requests.
To me, fly.io is an alternative to Heroku as the title of the thread is "on: Heroku Free Alternatives". So I was comparing it to Heroku and their pricing section UX.
Dude, I didn't mean it like that. Again, I am sorry.
My cloud experience is very limited. I have seen people talk about your SQLite related blog posts on HN, but I didn't participate in those threads, because I simply wasn't passionate enough or had an understanding about those topics.
I, too, love my job. But sometimes people with limited exposure to our product say some things about that mildly annoys me. But at the end, they don't know any better. I have unfortunately become one of those guys in this circumstance here. To me, the value proposition of a PAAS is only limited to the free pricing tier. It is a very insignificant comment about a UX thing.
I wish I could delete that comment. Please, I hope you take no further offense.
I've used a lot of the alternatives listed here but each one had some drawback or another. In contrast, I got a cheap 5 dollar Hetzner server and it's more powerful than any of what the free options here give you (Hetzner gets you 2 AMD vCPUs and 2 GB RAM), plus unlike AWS I never have to worry about whether I'm gonna randomly pay $10k this month due to a traffic spike. The only thing that was missing before was a good PaaS solution for the server (and I used to use Dokku before primarily) but Coolify solves that neatly.
They might be fun for random exploration if you're killing time, but not really a tool
Of course, you could get that same discoverability from language-package manager tool/website, if such systems ever allowed users (not developers!) to tag or otherwise multiply categorize packages, the way that e.g. Spotify allows users to create discoverable public playlists.
If you're already aware of all that, sure, they don't add much information. Concise comparison is pretty difficult...except in cases like this.
I don't know if I agree. I think maybe thorough comparison is difficult. But if you have a list of, for example, open-source alternatives to Postman, just including something like the amount of GitHub stars a project has can go a really long way. Especially if it becomes a list of 20+ projects and many of them have less than 10 stars or are no longer being developed
The difference is basically the difference between a list and a table.
Huge bonus if you use something like a spreadsheet software to make it interactive/sortable
Put it in an airtable. Done.
https://github.com/agamm/awesome-developer-first#deployment-...
I might add the limits too - a cool idea.
For reference, my use case/requirements:
- I want to be able to deploy dockerized apps to an AWS region.
- I should be able to define different process types with different env vars/instance scale
- I don't want to have to worry at all about underlying infrastructure
- I want to be able to pipe the logs out to some other unified logging
- I want to be able to deploy from a CI pipeline.
So far nothing else I've seen besides heroku fits the bill.Then I tried Django, and all hell broke loose... Still some ways to go to catch up with Heroku on all fronts I suppose, but for Rails, clear recommendation.
Sorry to hear Django was a pain. There’s a guide in the works for launching Django apps, but I can’t remember the exact status of that. https://community.fly.io/ is a good place to search for others who migrated or ask for help on there.
Also, dokku author is very responsive in GitHub issues list.
Isn’t that what AWS Elastic Beanstalk is about? AWS supports a wide variety of different approaches.
I was trying something on a simple Django project that nobody else accessed except me for like 3 times. AWS charged me $30 that month. Nobody knows what load they were balancing since I barely used the app.
I shut down everything.
I know Elastic Beanstalk and Lightsail both exist; but they’re both much more complicated than `git push heroku master`. And that’s not even counting what you have to do/learn to be able to wire up a DB or Redis instance or Email deliverability on AWS.
Straightforward to you may still be magic to most non-tech people. They will pay an insane premium to not have to worry about this type of stuff.
And to do the setup of a bare metal web service was not much more to know.
This is especially true looking at prebaked stuff like what digital ocean offers. You can use their Django image and the entire thing is ready to go.
Ultimately, docker and docker orchestrations are fairly standard devops skills at this point.
I think these paas solutions take devs who might otherwise bite the bullet and learn how to work with these in the wrong direction.
that said, perhaps not everyone is a sysadmin and security expert. In fact, this was the whole beauty of Heroku; "git push heroku" and you're done.
In my eyes, a rented server is a rented server regardless of which data center or network provider is behind it. It's not emotional, merely an optimization game.
Most of their assholery is towards employees, suppliers, and competitors - rather than customers.
Oracle, on the other hand, has a reputation for extracting maximum money from locked-in customers; an aggressive legal team; and producing products like Oracle Financials which are widely hated by users.
With that said, it's generally good advice to not count or rely on anything free for important things.
Hence the motivation behind this story; Heroku used to be free and now it's evaporating away back into the ether.
My only goal here is to provide helpful information. Cheers.
I will pay more to avoid Oracle.
> AWS Cloud Development Kit (CDK) is an open-source software development framework used to model and provision your cloud application resources with familiar programming languages. AWS Cloud Development Kit (CDK) accelerates cloud development using common programming languages to model your applications.
If you have used or read about Hashicorp Terraform, this is basically the same thing but limited to AWS. The “Getting Started” website shows a bit of code if you want a preview — https://docs.aws.amazon.com/cdk/v2/guide/getting_started.htm...
It was a horrible experience.
Note: I'm the maintainer of Dokku.
Although the performance was abysmal. It’s either their compute or postgres, but my import of 25mb json via a script took 22 minutes ($450/month) on job plan, while my 2GB digital ocean box did the same in 3 minutes.
Also email support takes 3 business days to respond.
But render has great potential. If they fix the performance and straighten out the support, they would definitely win over a lot of heroku customers.
Probably will just migrate to Dokky and be done with it.
The submission contains a lot of interesting comments on the perceived pros and cons of using a service like Fly.io from a variety of viewpoints.
I'm happy to also answer any questions here about my DO k8s -> Fly.io migration experience.
Currently I have a directus app with sqlite and litestream on it but will probably move away from that to something custom as I don't have a good story regarding migration and ci/cd.
https://www.heroku.com/free
https://blog.heroku.com/next-chapterJust wanted to give a shoutout to Netlify, great free tier and very easy to use
Anyone have better options than a droplet? My budget is around $5/month ideally
(They're small projects, so not running into paid tier yet)
But, they also have had little innovation in years, anecdotally the rate of issues is increasing, they released a very uninteresting roadmap, and the future level of investment from salesforce seems iffy.
Nothing stops you from using cron in your Dockerfile:
RUN apt-get install -y cron
Actually I just did it a few weeks ago and it works as expected.fly.io is universal, it just runs your Dockerfile. Everything else is up to you.
It's weird, because there's a group of (understandably) devoted Heroku users who have very high standards for fidelity to Heroku's DX, and while we're thrilled to do our best to meet (or at least approach) those standards, being Heroku is actually not the primary mission of our team; getting people's apps running quickly and close to their users globally is.
It only supports hourly, daily, or every 10 minutes and it isn't even guaranteed to run. I would see it skip jobs more often than I expected, maybe 1 in 20 at times.
Supporting actual cron is way better than anything Heroku offered.
Don't need to 'rail against' someone building a product that doesn't exactly fit your needs.
So, here is what I did. Added a simple API layer in front of my script. The API layer takes in a secret message to a specific endpoint that initiates a background process. Initially, I used my raspberry pi to periodically make the API requests. But I plan to try out: https://cron-job.org/en/
Having heard about fly.io from fasterthanlime I was expecting things to be smoother but there were a few gotchas.
Eventually I managed to configure things in the right way but it was a far cry from heroku buildpacks.
I'm trying to get my clients (1M+ spending per year) to switch from azure (absolute dumpster fire) to fly.io but I'm not sure it's mature enough just yet.
At that point I'd rather have to configure standard machines on hetzner than finding the quirks of a random paas
The best aspects of fly.io are 1) effortless multi-regional deployments 2) technological freedom 3) price. For example: Azure = $140, Fly.io = $10 for the same deployment. We are in love with those guys and gals.
But it's not all sunshine and roses. Some fly.io users have intermittent problems during deployments (see forum). I believe those issues will be rectified and improved with time.