Overview
Lucene or Apache Lucene is an open-source Java library used as a search engine. OpenSearch is built on top of Lucene.
OpenSearch converts Lucene into a distributed system/search engine for scaling horizontally. OpenSearch also provides other features like thread-pool, queues, node/cluster monitoring API, data monitoring API, Cluster management, etc. In short, OpenSearch extends Lucene and provides additional features beyond it.
OpenSearch hosts data on data nodes. Each data node hosts one or more indices, and each index is divided into shards with each shard holding part of the index’s data. Each shard created in OpenSearch is a separate Lucene instance or process.
Notes and good things to know
When an index is created in OpenSearch, it is divided into one or more primary shards for scaling the data and splitting it into multiple nodes/instances.
- As each shard is a separate instance of Lucene, creating too many shards will consume unnecessary resources and damage performance.
It takes proper planning to decide the number of primary shards for your index, taking into account the index size, max growth, and the number of data nodes.
- Previous versions of OpenSearch defaulted to creating five shards per index. Starting with 7.0.0, the default is now one shard per index.
