search-systems.md

Various search databases and backends as alternatives to Elasticsearch.

Rust

• Sonic - github.com/valeriansaliou/sonic - lightweight & schema-less search backend
• Tantivy - github.com/quickwit-oss/tantivy - full-text search engine library inspired by Apache Lucene
• Toshi - github.com/toshi-search/Toshi - search engine built on top of Tantivy
• Bayard - github.com/mosuka/bayard - search engine built on top of Tantivy
• Meilisearch - github.com/meilisearch/meilisearch - fast and full-featured search engine

C++

• Xapiand - github.com/Kronuz/Xapiand - restful search engine
• Typesense - github.com/typesense/typesense - fast, typo-tolerant, fuzzy search engine
• Manticore - github.com/manticoresoftware/manticoresearch - search database

Go

• Bleve - github.com/blevesearch/bleve - modern text indexing library
• Blast - github.com/mosuka/blast - search engine built on top of Bleve
• Bluge - github.com/blugelabs/bluge - newer indexing library originally evolved from Bleve
• Zinc - github.com/zinclabs/zincsearch - search engine built on top of Bluge
• Phalanx - github.com/mosuka/phalanx - search engine built on top of Bluge with clustering

Javascript

• Minisearch - github.com/lucaong/minisearch - search engine for browser/node

• C++ - Manticore Search - https://github.com/manticoresoftware/manticoresearch

If I may add more informations on the Rust alternatives:

• Tantivy & Bleve are alternatives to Lucene, the library that Elasticsearch and Solr use under the hood
• Toshi is to Tantivy what Elasticsearch is to Lucene
• Bayard provides RAFT consensus (for replication) and a gRPC interface on top of Tantivy
• MeiliSearch focus end-user search, site-search or app-search, a subset of what you could build with ES

Would it be possible to add LambdaDB a Go in memory database for searching.
Lambdadb makes it possible that with just a csv of the data, into having contains search up and running within minutes.

It's used in production at multiple companies, a example of the database with +9 million records is running here, click the yellow.

There are even some instances with +100 million records.

Although is used, there haven't been many feature requests lately.