With Heroku there is always a ‘but’. For 15 years now, I’ve heard Heroku described as “magical,” the gold standard for developer experiences. However, Heroku doesn’t seem to be able to live up to his mythology. That’s not to say Heroku hasn’t had a significant impact on other services and products — but there’s a reason why Kubernetes, rather than Heroku, is becoming the standard when it comes to building and scaling applications. Some say Heroku was ahead of his time. This can be. Or maybe the magical developer experience was simply priced too high to function in the modern clutter of enterprise computing.

Because the company recently scrapped its free offer has announcedHeroku is back in the headlines. The reason: It was too much work to stop the abuse of his free offer. “Our product, technical and security teams are working extremely hard to monitor fraud and abuse of the free Heroku product plans,” said Bob Wise, Heroku general manager and executive vice president at Salesforce (which Heroku acquired in late 2010).

Rather than playing cat and mouse with crypto scammers, the company hopes its investments will be better off with its customers — who are likely to be less than they should be. That may sound like criticism, but it’s not meant to be. In my community, Heroku is only praised for the way it has revolutionized the way applications are delivered. This lasted just as long – if not longer – than the creation before Heroku.

The problem, according to Jason Warner, lead developer at Heroku between 2014 and 2017, was that “Heroku was never finished.” The acquisition by Salesforce was there not helpful – this put Heroku’s development on hold. As the ex-lead developer puts it, “Heroku was magical for some uses. A completed Heroku could have been magical for many more uses.”

Despite Heroku’s impact on the industry, it appears that the company has fallen into a black hole at Salesforce.

“I don’t know exactly, but I suspect that Heroku has fewer employees today than five years ago,” tweeted Craig Kerstiens, product lead at Crunchy Data and former Heroku employee. That’s surprising, considering that there are more applications running on Heroku today than five years ago. It’s likely that Heroku has always been somewhat orthogonal to Salesforce’s core business, making it difficult for the company to get the internal funding it needs. As an original but closed developer experience, Heroku could never be more than Salesforce wanted. There was no opportunity for outside investment, which became a growing problem as community involvement with Kubernetes grew. Everything could have gone very differently, says developer Scott Williams, for example:

Very yes. It had a tone of potential and was more intuitive than openshift v2. If it had become open source around the time Kubernetes came out, we would probably all still be taking it seriously.

-Scott Williams?? (@vwbusguy) August 27, 2022

Google’s James Ward sees the biggest problem elsewhere:

I believe by far the biggest problem was the lack of an escape hatch. When a team/organization reached the limits of Heroku, the only option was to do all the things Heroku offered – infra automation, ops, etc. With Kubernetes, there are useful abstractions and escape hatches.

— James Ward (@_JamesWard) August 27, 2022

Therefore, the developer Jeremy Chone is probably right when he writes:

Equivalent to #GoogleAppEngine. PaaS makes it easy to start projects and difficult to complete.

Kubernetes has hit it well.#heroku #Easter #ClupApp #AWS #kubernetes

— Jeremy Chone (@jeremychone) August 27, 2022

Enterprises tend to rely on chaotic magic: an enterprise developer might like a PaaS that is highly idiosyncratic and makes building a specific class of applications incredibly easy. However, an enterprise development team needs to think about extending and adapting tools to the chaotic realities of existing and future infrastructure. (FM)

This post is based on an article from our US sister publication Infoworld.