Description
A database LIKE query starts crawling around a hundred thousand rows, and once you also want tokenizing, highlighting and per-category counts, you are hand-rolling everything. Apache Solr is the search server built for exactly that: feed it documents to index, then run full-text search, filters and faceted counts over plain HTTP, with answers in milliseconds.
It is built on Lucene and scales out with shards and replicas, so a node dying does not stop queries. The bundled admin console lets you try queries, inspect cores and read JVM metrics in the browser, so tuning search relevance takes no code.
Full-text search: Tokenizing, relevance ranking, highlighting, spell correction and synonyms: most of what a search box needs.
Facets and filters: Count hits per field value, so product filters and category counts need no extra query.
Vector and geospatial search: Beyond keywords, it handles vector similarity search and coordinate-based geographic queries.
REST API: Indexing and querying go over HTTP with JSON or XML, so any language can connect.
Distributed and fault tolerant: Collections can be sharded with several replicas, and replication and failover are handled by the cluster.
Admin console and monitoring: The web UI runs queries and shows core and collection status, while JMX metrics plug into monitoring.
It is built on Lucene and scales out with shards and replicas, so a node dying does not stop queries. The bundled admin console lets you try queries, inspect cores and read JVM metrics in the browser, so tuning search relevance takes no code.
Features
Full-text search: Tokenizing, relevance ranking, highlighting, spell correction and synonyms: most of what a search box needs.
Facets and filters: Count hits per field value, so product filters and category counts need no extra query.
Vector and geospatial search: Beyond keywords, it handles vector similarity search and coordinate-based geographic queries.
REST API: Indexing and querying go over HTTP with JSON or XML, so any language can connect.
Distributed and fault tolerant: Collections can be sharded with several replicas, and replication and failover are handled by the cluster.
Admin console and monitoring: The web UI runs queries and shows core and collection status, while JMX metrics plug into monitoring.
Tags:search-engine