Managing 10TB of Data in Elasticsearch with 30GB Heap: Best Practices and Considerations

I was very interested in frozen indices, but it seems like they are deprecated.

According to the blog, You would now only need 2.5GB of heap memory to keep 10TB of geonames-like data indices open.

  1. I want to understand if I am handling 10TB data in one machine with a 30GB heap, and what consideration I need to take care of so as not to break the node considering I have the necessary resources. I can throttle the queries sent to ES based on the number of indices and count of queries.

  2. Once the query sent to ES is completed, will ES free up the memory taken up to serve the Query immediately or will it be freed after some time?