We are interestead on using Elasticsearch in our production servers for indexing JBoss logs. My data estimation is about 26Gb/day every month day, except last month day where the data spikes up to 45Gb/day . Our data retantion period will be 10 days. I also estimate that we wont do more than 50 ~ 70 queries per day.
I was thinking a bit about the hardware requirements. The topology will include 3 master nodes with:
16GB of ram
6-8 Core cpu's
350 Gb of HDD
I was thinking that since we wont do many queries and we wont retain data for a long time, data nodes may not be necessary.
I would like to hear your thoughts and advice