The cloud may be a prison, but its infra services, specifically EC2+Autoscaling Group, S3, SQS, and DDB, make the cloud a Four-Seasons-like prison. So, what spe

The cloud may be a prison, but its infra services, specifically EC2+Autoscaling Group, S3, SQS, and DDB, make the cloud a Four-Seasons-like prison. So, what specifically can we do to make local-first at least equality productive for a med-sized company like Uber for at least the following simple (doesn't mean they are easy, though) requirements: - Anyone in the company can launch and configure an autoscaled VM cluster. It's surprisingly hard to create a system like EC2, as Uber failed to do so even though they tried, and a number of multi-billion-dollar internet company tried and also failed. - Anyone can store any amount of data and scan them with practically unlimited bandwidth for analytics. In fact, object storages on cloud have been so good that I failed to see a thriving open-source community that actively work on an alternative. - A database that just works and can handle a wide range of workload, including seemingly anti-patterns like using database as a large-scale distributed queue. Yes, I'm talking about DDB. - A simple queue that supports competitive consumer pattern and practically scales infinitely. Per Tim Bray, a queue that "is almost perfect" as SQS.