I've been intrigued recently by the microservices architecture idea, so I was trying to restructure an app I've written into this kind of a framework. My first step with this is to get the MongoDB database up and running in Docker. So I pulled the 'mongo' image for Windows…and it's 11 GB. I spawn a container, and it uses 872 MB of RAM.
Meanwhile, installing a MongoDB server directly on my machine uses about 1.6 GB on disk, and 67 MB RAM when running as a service. If I have each piece of my app running in a different container (orchestrated by Kubernetes), then the resource usage is going to be insanely high.
I read about how excited people are that they "don't have to install MongoDB directly on their machine," but why? This massive over-usage of resources upon dockerizing the service has me rethinking this whole plan.
What am I missing here?