Questions about ElasticSearch memory improvement

Hi,
i've been researching ElasticSearch's improvement in terms of memory usage.

Today, I've read this article(https://www.elastic.co/kr/blog/significantly-decrease-your-elasticsearch-heap-memory-usage). 

This article says that ElasticSearch's memory usage have been reduced because some data structure moved to Disk and it is on filesystem cache.

Then I've checked ElasticSearch 7.7 release notes to verify above is true or not.

However, the release note(https://www.elastic.co/guide/en/elasticsearch/reference/7.x/release-notes-7.7.0.html) says that data is moved onto off-heap instead of disk and filesystem cache.

I don't know which one is true and how can be memory usage reduced because of off-heap implementation, please help me.

Heap usage was reduced in Elasticsearch 7.9 l believe as more data was moved off-heap. I would therefore recommend upgrading to/deploying the latest version.

But, off-heap memory takes up memory too. I don't know why it is possible to improve performance.

it is juts because off-heap doesn't subject to Garbage Collection?

It is recommended to keep the heap size below 32GB, which has traditionally limited how much data a node can hold. Moving data off-heap allows smaller heaps or more data which is more efficient.

1 Like

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.