manticoresoftware/manticoresearch

Manticore Search is an easy-to-use, open-source, and fast database designed for search. It is a great alternative to Elasticsearch.

What distinguishes Manticore from other solutions is:

• It's very fast and therefore more cost-efficient than alternatives. In the current reproducible benchmarks, Manticore Search 27.1.5 is:

  • 340x faster than MySQL 9.7.1 and 6.51x faster than Typesense 27.1 for 1.1M Hacker News comments
  • 3.85x faster than tuned Elasticsearch 9.4.3 for 100M+ Hacker News comments
  • 5.03x faster than tuned Elasticsearch 9.4.3 for selected typical DevOps queries on 10M Nginx logs; 1.71x faster than ClickHouse 26.6.1.1193 across the dashboard's default 10M-log query selection
  • 2.02x faster than tuned Elasticsearch 9.4.3 and 3.16x faster than ClickHouse 26.6.1.1193 for 1.7B NYC taxi rides
  • For the same 10M Nginx-log ingestion, Manticore Search Columnar 27.1.5 completed in 5m 46s vs 10m 15s for tuned Elasticsearch 9.4.3, using 1.02 vs 3.80 CPU cores on average, 3.98 GB vs 36.98 GB RAM on average, and 0.41 MB read / 8.05 GB written vs 322.14 MB read / 18.47 GB written.

  Results are workload-specific; use the linked dashboard to select the queries that match your workload.

• ⚡ Multi-threaded query execution and efficient query parallelization use all CPU cores for low response times.

• 🔎 Full-text search works seamlessly with both small and large datasets.

• 🧩 Hybrid search combines full-text and vector retrieval in a single query for better relevance.

• 🧠 Auto-embeddings with automatic chunking generate vectors from text and can split long documents into fixed-size, recursive, or sentence-based chunks stored in a float_vector_array, making relevant sections searchable without application-side chunking.

• 💬 Conversational search lets existing vectorized tables answer questions with KNN retrieval, conversation history, and LLM-backed responses through SQL CALL CHAT or the HTTP JSON /search endpoint.

• 🏎️ Row-wise storage provides fast performance for small, medium, and large datasets.

• 🗄️ Columnar storage, powered by the Manticore Columnar Library, handles datasets too large to fit in RAM.

• 📇 Automatic secondary indexes use the PGM-index (Piecewise Geometric Model index) for efficient mapping between indexed keys and memory locations.

• 🧠 Cost-based query optimization uses indexed-data statistics to select efficient execution plans.

• 🧮 SQL-first and MySQL-compatible: use SQL as the native syntax with your preferred MySQL client.

• 🔌 Official clients for PHP, Python, Python asyncio JavaScript, Typescript, Java, Elixir, Go, and Rust make integration easy.

• 🌐 HTTP JSON API and Elasticsearch-compatible writes support versatile data and schema management.

• 🧱 Sharded tables transparently distribute reads and writes across physical shards on one node or across a replication cluster.

• 🔐 Built-in authentication and authorization secure MySQL, HTTP/HTTPS, distributed remote agents, and replication-related operations with users, bearer tokens, and fine-grained permissions.

• 🪶 C++ performance and a small footprint: Manticore Search starts quickly and uses around 40MB RSS for an empty instance.

• ⚡ Real-time inserts make newly added or updated documents immediately accessible.

• 🎓 Interactive courses at play.manticoresearch.com make learning easier.

• 🔄 Virtually synchronous multi-master replication uses Galera and includes load balancing.

• 🔗 Data synchronization supports sources such as MySQL, PostgreSQL, ODBC, XML, and CSV.

• 🛡️ Safe writes: while not fully ACID-compliant, Manticore supports isolated transactions and binary logging.

• 💾 Backup and recovery include manticore-backup, SQL BACKUP, and S3-compatible backup/restore support.

Craigslist, Socialgist, PubChem, Rozetka and many others use Manticore for efficient searching and stream filtering.

• Full-text search and relevance:
  • Over 20 full-text operators and over 20 ranking factors
  • Custom ranking
• Other search capabilities:
  • Rich filtering functionality
  • Fuzzy search
  • Faceted search
  • Geo-spatial search
  • Hybrid search
  • Vector search
  • Auto-embeddings with automatic chunking
  • Conversational search
  • Joining tables
  • Spelling correction
  • Autocomplete
  • A wide range of functions for filtering and data manipulation
• Natural language processing (NLP):
  • Stemming
  • Lemmatization
  • Stopwords
  • Synonyms
  • Wordforms
  • Advanced tokenization at character and word level
  • Proper Chinese segmentation
  • Text highlighting
• Stream filtering:
  • using a "percolate" table
  • or the Kafka integration
• High-availability:
  • Sharded tables
  • Data can be distributed across servers and data-centers
  • Synchronous replication
  • Built-in load balancing
• Security:
  • Authentication and authorization
  • https support
  • read-only mode
• Data safety:
  • manticore-backup tool and SQL command BACKUP to back up and restore your data
  • S3-compatible backup and restore
• Data storages:
  • row-wise - requires more RAM, provides faster performance
  • columnar - requires less RAM, still provides decent performance, but lower than the row-wise storage for some kinds of queries
  • docstore - doesn't require RAM at all, but allows only fetching original value, not sorting/grouping/filtering
• Performance optimizations:
  • Secondary indexes
  • Cost-based optimizer determines the most efficient execution plan of a search query
• Data types:
  • full-text field - inverted index
  • UUID document IDs for real-time tables
  • int, bigint and float numeric fields in row-wise and columnar fashion
  • multi-value attributes (array)
  • string and JSON
  • on-disk "stored" for key-value purpose
• Integrations:
  • Sync from MySQL and PostgreSQL
  • Sync from XML
  • Sync from CSV
  • Sync from ODBC
  • Sync from MS SQL
  • Sync from Kafka
  • With MySQL as a storage engine
  • With MySQL via FEDERATED engine
  • ProxySQL
  • Apache Superset
  • Grafana
  • Fluentbit
  • Kibana (Demo)
  • Logstash/Filebeat
  • Vector.dev
  • Mysqldump
  • Manticore Columnar Library

For Linux and macOS with Homebrew, use the one-line installer:

curl https://manticoresearch.com | sh

It detects Debian/Ubuntu/Mint, RHEL/CentOS/Amazon/Oracle/Fedora-like RPM systems, and macOS with Homebrew. It configures the official Manticore repository when needed, installs manticore, starts the service when supported, and prints the result.

Common commands:

curl https://manticoresearch.com | sh -s help
curl https://manticoresearch.com | sh -s list-versions
curl https://manticoresearch.com | sh -s version 25.0.0
curl https://manticoresearch.com | sh -s dev
curl https://manticoresearch.com | sh -s upgrade

Docker image is available on Docker Hub.

To experiment with Manticore Search in Docker just run:

docker run --name manticore --rm -d manticoresearch/manticore && until docker logs manticore 2>&1 | grep -q "accepting connections"; do sleep 1; done && docker exec -it manticore mysql && docker stop manticore

You can then: create a table, add data and run searches. For example:

create table movies(title text, year int) morphology='stem_en' html_strip='1' stopwords='en';

insert into movies(title, year) values ('The Seven Samurai', 1954), ('Bonnie and Clyde', 1954), ('Reservoir Dogs', 1992), ('Airplane!', 1980), ('Raging Bull', 1980), ('Groundhog Day', 1993), ('<a href="http://google.com/">Jurassic Park</a>', 1993), ('Ferris Bueller\'s Day Off', 1986);

select highlight(), year from movies where match('the dog');

select highlight(), year from movies where match('days') facet year;

select * from movies where match('google');

Note that upon exiting the MySQL client, the Manticore container will be stopped and removed, resulting in no saved data, so use this way only for testing / sandboxing purposes.

Read the full instruction for the docker image for more details including our recommendations on running it in production.

If you prefer explicit package-manager commands, use the manual instructions below or see the downloads page.

sudo yum install https://repo.manticoresearch.com/manticore-repo.noarch.rpm
sudo yum install manticore

or use dnf when available:

sudo dnf install https://repo.manticoresearch.com/manticore-repo.noarch.rpm
sudo dnf install manticore

wget https://repo.manticoresearch.com/manticore-repo.noarch.deb
sudo apt install ./manticore-repo.noarch.deb
sudo apt update
sudo apt install manticore

brew install manticoresoftware/tap/manticoresearch manticoresoftware/tap/manticore-extra

For WSL/WSL2, use the Linux installer inside the WSL distribution. For native Windows, see the Windows installation instructions.

• Documentation
• Interactive courses
• Manticore Community Forum
• Public Slack chat
• Public Telegram chat (En)
• Public Telegram chat (Ru)
• Bug tracker

• https://dev.tiki.org/Manticore-Search - TikiWiki CMS Groupware integration
• https://github.com/EvilFreelancer/laravel-manticoresearch - Laravel ManticoreSearch plugin
• https://www.mediawiki.org/wiki/Extension:SphinxSearch - extension for MediaWiki
• https://github.com/gordonbanderson/silverstripe-manticore-search - integration for SilverStripe
• https://drwho.virtadpt.net/archive/2022-05-13/combining-manticore-and-searx/ - integration for SearX
• https://github.com/dimv36/sphinxlink - Extension for PostgreSQL for running search queries and returning data as PostgreSQL tables

Should your company require any help - we provide full-cycle services:

• Audit
• Support
• Consulting
• Development
• Training

More details here

Manticore Search is an Open Source project with development made possible by support from our core team, contributors, and sponsors. Building premium Open Source software is not easy. If you would like to make sure Manticore Search stays free, here is how you can help the project:

• Donation through PayPal
• Become our client and let us help you

Manticore Search is distributed under GPLv3 or later. Manticore Search uses and re-distributes other open-source components. Please check the component licenses directory for details.