Yeah, I moved a while back to DO's App Platform. I've not had any issues with them so far. Recommend them a lot.

I am considering DO App Platform for a new project. Would you be open to sharing any lessons learned ?

This is surprising to me. I actively used Heroku during the early to mid 2010s. I do not remember ever seeing the Salesforce logo there much later.

Salesforce mostly left Heroku to do its own thing for a long time. Since that changed, dumpster fire.

I think they’d be happy if all the customers moved on. They just don’t want to upset enterprise customers.

> lead to customers moving elsewhere

Since they're no longer accepting new enterprise clients, maybe this is intentional.

Yeah I agree. I saw someone say that they were not in KTLO mode, but this seems pretty bad, especially considering it happening at Salesforce.

Often seems like there's no defeat that Benioff can't steal from the jaws of Victory.

I really like Railway, and have deployed many sites with them, but got worried by their recent funding round. At some point those investment bills are going to come due.

Nothing to study. A common scenario when a mega corp acquires an incredibly successful startup and then lets it die. Happens more often than not. This is why I chuckle when I see an acquisition and the founders claim "Nothing is changing. We are not going anywhere" . There may be exceptions but the moment a hugely successful company like heroku gets acquired, you know it's most likely game over. To their credit, they survived 15 years after acquisition but barely.

I remember feeling the same way about Slicehost back in the day after Rackspace acquired them. Loved Slicehost. Not too long after though, Digital Ocean appeared with everything I loved about Slicehost and has kept getting better ever since.

I feel like that's Fly.io now. They took all of the great things about Heroku but also dramatically improved and added new capabilities...while improving on pricing, particularly for lower traffic stuff. Love Fly.

What is there to be studied? Once a company is acquired you bounce. There is usually a two year grace period before you start feeling the pain as a customer, which should give you the time to migrate.

It is a typical acquisition by the book, always goes the same way after three to five years.

> Salesforce acquired them and just let it die, baffling.

This is a common misconception, but it's actually not true. The reality is even more bizarre.

Most of Heroku's successful years came after the acquisition, not before. Heroku was acquired extremely early in its lifecycle, and Salesforce does actually bear responsibility for investing in it and making it the powerhouse it became. Most of what people remember as the glory days of Heroku came long after the acquisition. And in fact, at the time of acquisiton, Heroku was nowhere near as competitive as a product as it later became.

It was only much later on that Salesforce began to pull the supports out from underneath it, leaving it to fall behind and become what it is today.

The narrative of "BigCo™ acquires startup, then leaves it to wither and die" is a trope because it is very commonly true, but it's actually not what happened in this particular case.

Amongst the people opting for just plain Linux servers Linode was the big name back then. They later got supplanted by DigitalOcean, and both are of course also run into the ground by now.

Downfall? The founders and VCs made tens of millions of dollars. That’s the success condition for them.

They are discontinuing it, you were right.

Dogfooding their future products!

I've been very happy with https://render.com/ , seems the closest to what heroku was

Fly.io is incredible.

What is the concern exactly? (Product/platform enshittification?)

Dokku. The self hosted alternative.

As a former enterprise person, this clearly states “exiting growth cycle into low-staffing maintenance mode”; Salesforce must have bought them to kill a price-beating competitor to multi-year Salesforce PaaS contracts, same as Okta did with Auth0. Investors are typically-majority short-sighted and only care about growth-cycle revenue, so once they reached market saturation, they were ripe and duly reaped. So long, Heroku.

"sustaining engineering model"

ie, life support.. bit rot will set in, they are dead.

Oh, they're very clear, just not explicit.

Holy crap is this underselling how poorly this announcement is structured. Not only does it not provide clarity, it words things in such a way that it just begs more questions. “There are no changes for now”....

In a company I used to work for, "sustaining engineering" was the team of developers that handled all of the bugs and issues reported by customers on old-but-still-technically-supported versions of the products. (The ones who worked on current versions of the products where just "engineering.")

So basically heroku will fix whatever is broken, but don't expect any new features or development.

We've been optimizing for decades to engineer the bullshit-generating super-soldiers required to craft modern PR statements.

I found myself wondering where the rest of it went.

It's like PBS, they are going to beg for your money now with a sustaining engineering membership

Not sure why people love fly.io over all the other competitors so much. I myself prefer render.com, for the simplicity and predictability of their billing, and their deployment model is so intuitive

I moved to render years ago and have been very happy with the decision. It feels like heroku, if it never got acquired by salesforce and kept improving.

It was the easiest place to host my Rails apps back in the day.

Yes, Render if you want something similar.

yes,, render feels like the most natural next step right now (similar mental model). Still kind of nostalgic about Heroku, had really good times with it.

Didn't you read the last line of the announcement? They have enterprise AI dollars to chase! Salesforce wants some of them billions. Gotta make up for the 42% their stock price has dipped in the last 12 months.

Exactly this. Missed opportunity.

It would be impossible to buy a company as a steward. We worked with them. In the early 2020s, their teams were as lazy as IBM boomers. Happy to hop on a call, but let's do an rain check or whatever and come back to you after 17 weeks to say no in an email CCed with 42 people playing hot potato between commercial people who try their best to sound technical. You need a mind map to know which person you were talking to and who introduced you to them.

My money is on holding companies like IAC buying the brand first through financial engineering and restructuring finances initially. They would load it up with debt like they did with the sporting goods store in Sopranos.

Afterward, they would sell it to a Euro-based caretaking company like Bending Spoons, with a focus on maintenance engineering rather than innovation engineering.

northflank comes close

If any Heroku customer is reading this and not immediately going "we need to move off Heroku ASAP" all future problems are their own fault.

It means: go elsewhere, they're dead.

> including helping organizations build and deploy enterprise-grade AI in a secure and trusted way.

Baffling

From a business perspective, this means they will not be investing in innovation on the platform anymore. Instead, they will focus their efforts on maintaining the current operations and keeping the lights on.

I am sure that the guy whose name is on the Post didn't even write this himself. Probably some corporate writer using LLM :).

Could have just said we will Keep the Lights on

Surely it's a typo and they meant "sustainable"?

Otherwise IMO such an odd word choice. Definition:

>> providing physical or mental strength or support

Often "nothing is changing" ends up being more literal than the founders realize or intend. Acquisitions by big companies tend to slow the development to a crawl as development bureaucracy takes over. When a great product is practically frozen in time it stops being great in 5-15 years as the rest of the world passes them by.

> They took all of the great things about Heroku but also dramatically improved and added new capabilities

I also love Fly, but they were missing easy managed databases (which always seemed like the main reason to use Heroku to me). And now they have them they're very expensive (even compared to Heroku). Which is a shame because their compute is very cheap.

I was working for Slicehost at the time, we were a tiny team working our butts off in a loft office in downtown St. Louis, with a few remote employees.

To my understanding there was a runway-growth problem. Could the founders raise and spend (efficiently) enough money quickly enough to keep the business viable? It would be a big gamble and the alternatives were to shut down (no way!) or sell. So they sold.

Rackspace wanted to take Matt’s and Jason’s know how (plus customer base) and go big, really big! That defocused our efforts a bit, plus there were corporate integration headaches (though not too bad). Eventually Linode, already a competitor, and later Digital Ocean filled the void.

Translation: We're going to reassign the engineers into Salesforce AI.

It's such a shame, because they had one of the best services out there. Being able to push via Git and end up with a running deployment was a killer feature. It may not have been the first (Elastic Beanstalk was way older but when it first came out it was Java only iirc, ick) but it was incredibly popular.

Seeing them now chasing AI as a "me too" after being acquired by Salesforce just shows that huge companies will acquire something then sit on it for years and let it rot.

How does render compare to fly.io? Does anyone has experience running production rails apps on these?

Do you have any public docs on how y'all migrate customers out of Heroku Postgres without downtime?

Seems to be the sticking point for a lot of people, myself included.

Railway for backend APIs. Render for front-end apps. That's my current go-to.

Although I would consider, _when possible_, using Vercel or Netlify.

My understanding is that Heroku is just an AWS reseller. I don't know if there's a lot of value in a PaaS piggy-backing on another PaaS anymore. Especially for Salesforce.

I think there's a couple decent alternatives out there: https://alternativeto.net/software/heroku/

There are also a lot of cool "self-hosted Heroku" alternatives

- Coolify (PHP) (2020) https://github.com/coollabsio/coolify

- Dokku (Go) (2013) https://github.com/dokku/dokku

- Dokploy (TypeScript) (2024) https://github.com/Dokploy/dokploy

- CapRover (TypeScript) (2017) https://github.com/caprover/caprover

- Komodo (Rust) (2022) https://github.com/moghtech/komodo

Salesforce acquired Heroku 15 years ago.

It saves face with investors to say you're shuttering a product to focus on the hot new thing as a strategic decision than to say you're shuttering it because your actions have led it to be unviable.

That has been the case for a very long time at this point, the Salesforce acquisition was a death knell. The only stuff i have left on Heroku are zombie projects I don't care about.

From a business perspective the have brought it out to the field with a shotgun.

I would love to see the prompt that led to that word salad

Heroku has been running in this mode for a long time. The only difference is they made it official.

As opposed to the relentless innovation they have demonstrated in the past 5+ years? /s

Sustaining is used in Engineering to mean that it's now post-GA and there is no further development. The platform is not End of Life but there are no more features planned.

Sustaining as in sustaining their shareholders.

What's the best alternative?

This reads more like "we won't deliberately turn the lights off… but they're probably gonna break on their own eventually".

If there was one thing we would all decide differently here at Fly.io, like if you gave us a time machine, is how we did databases. Someday Kurt and I will write the post about how those decisions came to pass and how they played out.

We're doing Managed Postgres now (MPG), which is what we should have done to begin with, but it took us for-ev-er to get here.

I thought if you were all in on AI you wouldn’t need engineers - just swarms of agents doing everything. :)

Yea, this is accurate in my experience.

For what it's worth, I've been very happy using Crunchydata PostgreSQL with Fly.

Huh, so that's what they mean when using the word "we". "We" is not Heroku, it's Salesforce.

There's a ton of value right now in the existing customers. None of them want to put effort into migrating away from Heroku if they can avoid it.

I don't see AWS as a PaaS at all. Heroku's selling point has always been how it makes everything way easier.

And Heroku has stagnated for at least 13 of those years...

Thanks for the reply! Most of these (also on the alternativeto) are self-hosted, which is different from having heroku do it. Also most only support webservers, while the majority of our servers aren't web..

All I can say is thank you. I learned to manage servers because of Slicehost and the articles on it back then.

I remember being excited by the merger because well, Rackspace had such a fantastic reputation at the time. People still tell stories about their service. The Rackspace Cloud was just up against an absolute monster in AWS and never really became competitive.

hey!

northflank supports the same buildpacks that you run on Heroku, so it should be fairly straightforward.

we have these docs for a more detailed walkthrough:

1/ https://northflank.com/docs/v1/application/migrate-from-hero...

2/ https://northflank.com/blog/how-to-migrate-from-heroku-a-ste...

why split, you could use railway and render for both front end and back end

Yup, their Git Push Deployment was really a killer concept and a huge gateway for people just writing good apps not needing to care about infra and still being able to get a production-ready setup.

Why/when do you use Railway over Render?

Literally nobody who seriously uses Heroku wants to self-host their own Heroku.

Time flies. I do think the vision lasted pretty long post acquisition, maybe 5 years or so, but then the inevitable seemed... inevitable.

Digital Ocean App Engine has the easy setup and GUI management that made Heroku popular.

The lights will stay on until they burn out or the power goes off, or someone bumps the light switch or steals the light bulbs.

If you like VPS, Hetzner with Dokploy. It works great, the UI has essentially all the features of Fly or Render that you'd use for deployment, like preview build URLs and environments.

Build.io came out of this exact problem a few years ago (I joined in 25Q4) - trying to be what Heroku could have been if it had continued to evolve.

We offer the same default simplicity/speed, but with the ability to go deeper once teams hit scale, cost, or workflow limits. Plus a pricing model that stays understandable and improves as teams scale rather than punishing them for it.

Fair warning: the website is pretty light right now. It’s mostly a placeholder while we prep a broader push over the few months. Happy to answer questions here if helpful.

Moved from heroku to fly.io three years ago and I don’t regret it, great platform occasionally goes down and requires a bit of attention but the support forum is great

Looks like that still has downtime for a Postgres migration- you're suggesting going into maintenance mode and just doing a dump/restore. I've seen that take hours once you hit the terabyte scale, depending on hardware.

I've had pretty good luck setting up logical replication from Heroku to the new provider and having a 10-15 minute maintenance window to catch up once it's in sync. Might be worth considering.

You might also want to add a warning about Postgres versions. There's some old bugs around primary key hash functions that can cause corruption on a migration. I've seen it twice when migrating from Heroku to other vendors.

Thank you for the kind words, brought back some fun and interesting memories. I spent a lot of time helping to write and edit those articles, as did my coworkers, glad they helped you!

Sorry, but telling people to take a logical backup of their database, and then download it onto their local work station is insane for a production application. First, a logical backup at any decent scale will fail, and second, I don't even have enough local storage to do that -- even ignoring the compliance issues with downloading a full copy of production data onto a work station.

For a company like Northflank, I'd expect actual production-grade documentation for migrating, not instructions that are only applicable to a toy app.

Couldn’t agree more. That “git push and you’re live” moment removed a huge amount of accidental complexity, and it’s been the guiding experience behind what we’re building at Build.io.

What DO you want?

Sorry, I wasn't saying you should split, I wanted to say that depending on what type of apps you are more leaning into one makes a bit more sense than the other. Render with their own CDN is quite good for frontend apps. In comparison, the whole config and auto scalling/provisioning of Railway makes it easier for backend app.

Of course you can do both with both of these services.

When bandwidth matters, when you don't want to over or under provision, when you need multiple seats: if you make a project with a small team, Render is going to be quite expensive because of the cost of each seat while Railway offers unlimited seats for they paid plans. Just the whole pricing is different, I found myself more leaning into Railway when doing calculations.

We had an enterprise account on Heroku. We invested a little in Dokku and moved to self host it as of a few months ago.

We now have a kubernetes (k3s) backed Dokku self hosted on Hetzner. Significantly cheaper but pretty robust.

Just saying that it's not literally, but you are right, most people wouldn't be interested in self hosting.

I had an issue with one of my Sprites (Fly.io also runs sprites.dev) and the CEO responded to me personally in less than 10 minutes. They got it fixed quickly.

I was a free customer at the time. I pay for it happily now.

Not to be confused with builder.io, or worse, builder.ai

Fly.io are absolute G’s. The product is awesome and the tech blogs they write are fantastic.

It didn't seem quite as fire-and-forget as doing `Heroku create` when I tried to use it 3-4 years ago, especially the database setup. Do you use their Postgres offering?

Bucardo would be an option: https://www.porter.run/blog/migrating-postgres-from-heroku-t...

Do you care to show prices? The true benefit of heroku for me was really predictable pricing model. Build.io website doesn’t have it on mobile site at all. I don’t want to look at demo, i want to hook up my credit card, set a monthly budget and explore

Very close to the worst alternative for people who actually need Heroku, but it won't stop people from plugging it to death and back.

I agree, I wouldn't either. You can import directly via a DB to DB import in the platform without involving your laptop.

https://northflank.com/docs/v1/application/databases-and-per....

Some folks want to do that, others want to import a backup directly, some want to spawn a read replica and sync their DB. Different strokes for different folks, all supported on Northflank.

llIIllIIllIIl & runako give me an email on steven[at]build.io and I'll share. As mentioned, we stripped the site back while we overhaul and we certainly didn't expect this today!

FWIW it doesn't look like pricing or details of the service offerings are available on the desktop site, either.

Eh, no, depends on why you used Heroku in the first place. Way back when, I used it because the UI was dead simple and it Just Worked™. If I can replicate that with a VPS and have a good UI around it that takes care of everything, it's functionally the same to me.