Heap and direct memory size for my usecase

Hi.
Im trying to determine the amount of heap and direct memory I should give to the instances.

So I have a very simple use case with only about ~40 millions of events per month.

Im recording data into druid in batches with appx 2 minute intervals between batches.

Will such insignificant amount of data still require a serious heavy production cluster as described in here:

http://druid.io/docs/latest/configuration/production-cluster.html

Would I be able to get away with dockerizing everything and running on host with just 16Gb Ram?

Thx

JC, for such a small volume of data things should be fine. I am curious around your ingestion spec and why it is causing memory errors.

thx for helping. but I had to cancel this project and went with elk instead due to lack of time.