> To heroku's credit... I'm not sure if I can remember this kind of widespread outage ever happening before. Against heroku's side... I'm not sure what a heroku

> To heroku's credit... I'm not sure if I can remember this kind of widespread outage ever happening before. Against heroku's side... I'm not sure what a heroku-deploying customer would be expected, by heroku, to do, to avoid downtime here. This is the trade-off of lock-in into simple deployments. It's largely the same, EKS/Beanstalk/maybe lambda + RDS + S3... Azure has similar offerines, GCP has similar offerings, Heroku has similar offerines. You can get your app up, running, hosted for a dime - cost based on usage. And you can do a lot of stuff with very little operational time invested. That's why this is so great for small startups with 5 technical dudes. However, you can't get out. You're comitted to their abstractions for relational databases, deployments, DNS, loadbalancing, ... If you go ahead and do the things you need yourself based on simple linux VMs, you can have much more mobility across cloud providers. Give us a terraform endpoint and some kind of centos image and we can probably host there, speaking of work. However, that's overall an expensive and non-trivial thing. Suddenly, you're having so many problems heroku has a solution to. Suddenly you need someone who knows how to run a postgres cluster and how to handle failovers, or how not to. Suddenly you need someone who knows how drives should be handled. This is becoming a really big rift I'm overall starting to see. Applications and deployments are becoming more and more trivial, but there aren't many people understanding how to make the stack below it run reliably, with low chances of data loss.