Most of Heroku's revenue is from Postgres. They're arguably a managed Postgres provider with some app hosting tooling (this is not fair, but an interesting way to think about Heroku).
We're not Heroku. We _like_ how amazingly easy it is to get an app up and running on Heroku. But most of us were interested in flexing the underlying infrastructure to build new types of apps. So Heroku felt very constrained.
We see ourselves as, like, a good RPG. Quick to get going, rewarding and replayable when you go deep.
I'd dispute that MOST of the revenue is from Postgres, though it is likely a very large double digit percentage of revenue.
Their managed Postgres got an unfair advantage early on by being the default. With a new rails app you were just given a Heroku Postgres database. Heroku's Postgres offering was over 5 years ahead of RDS Postgres and a large part of the reason Amazon added support for Postgres for years they fought it, but eventually had to give into the constant customer request. Unfortunately that innovation of Postgres has stalled out a bit over time due to Salesforce starvation.
I do admit for what people think of as PaaS the Postgres revenue is significant, the part people would probably be surprised about is how much revenue the add-ons marketplace constitutes.
Interestingly New Relic got the same privileged experience in the early days and it resulted in the same type of revenue and mass adoption for New Relic.
We're not Heroku. We _like_ how amazingly easy it is to get an app up and running on Heroku. But most of us were interested in flexing the underlying infrastructure to build new types of apps. So Heroku felt very constrained.
We see ourselves as, like, a good RPG. Quick to get going, rewarding and replayable when you go deep.