[cholantesh]

Interesting; I may be biased because I've been involved in helping teams containerize as part of a cloud migration and only one or two cases has there been a real 'need', basically for running a Windows service that was eventually retired in favour of a lambda triggered by consuming a message in a queue.

[maccard]

We were waaaaay too big to fit in lambda layers. Our containers were 8GB when I left, and that was using all sorts of tricks on the host infra to share data between running containers.

The root of the problem was we had third party tools which were windows only.