It's really hard to build a system for the cloud that can run on bare metal unless you make that a goal of the project.
That can be a hard goal to sell to business. Especially when it requires more head count.
That can be a hard goal to sell to business. Especially when it requires more head count.
We see a trend today that many are using Kubernetes to build their platform, and that can run anywhere.
Unless you go all in on Lambdas and step functions, you are usually not that dependant on cloud (or K8s for that matter).
My workplace has simple services that are built as Docker images running on K8s, connects to Postgresql databases and sends messages using SQS. Only SQS is specific to AWS, and can be easily replaced with a different message queue (we are using Spring abstractions).
> Especially when it requires more head count.
Quote from the article:
"""The cloud was not able to make things easier to a point where we could measure any productivity gains,” he says, noting his operations team has always been about the same size."""
You still need an ops team once you scale up from a small startup and you don't necessarily end up with lower costs for the team itself. And if you take the added cost of the cloud infrastructure, it should in theory get harder to justify for companies as they grow. But I guess "Nobody ever got fired for choosing Cloud"
That’s an overly broad statement and likely false for many workloads .