HN
Today

Manticore Search 27.1.5: Auth, sharding, conversational and faster vector search

Manticore Search rolls out its 27.1.5 update, a major release consolidating features from previous versions. It introduces built-in authentication, sharded tables for easier scalability, and integrated conversational search capabilities. This release appeals to developers looking for robust, performant search and data management solutions with modern AI/ML features.

5
Score
0
Comments
#11
Highest Rank
6h
on Front Page
First Seen
Jun 22, 12:00 PM
Last Seen
Jun 22, 5:00 PM
Rank Over Time
221117212430

The Lowdown

Manticore Search has announced the release of version 27.1.5, which is a comprehensive update bundling changes from versions 25.0.1 through 27.1.5. This release focuses on significant new features, performance enhancements, and numerous bug fixes, solidifying Manticore's position as a powerful search engine. Key areas of improvement include security, scalability, AI-driven search, and overall system stability.

  • Authentication and Authorization: Manticore now includes native support for users, passwords, bearer tokens, and granular permissions across its various interfaces (MySQL, HTTP/HTTPS, remote agents, replication). This moves access control directly into the product.
  • Sharded Tables: The engine can now create, manage, and distribute inserts across sharded tables, simplifying operations for large, write-heavy deployments by centralizing sharding logic.
  • Conversational Search: A new feature allowing direct Q&A over vectorized data using CREATE CHAT MODEL and CALL CHAT. It leverages KNN on FLOAT_VECTOR fields, builds LLM context, and maintains conversation history.
  • Faster Vector Builds and KNN Improvements: Significant enhancements to vector search performance, including multithreaded HNSW builds, improved KNN distance calculations, and local ONNX embeddings support, leading to faster indexing and query times for large vector tables.
  • Better Faceting and Aggregations: Introduces facet_filter_mode for advanced e-commerce-style filtering and expands analytical capabilities with new statistical aggregations like percentiles and date_histogram with time zone support.
  • Other Notable Additions: Includes searchd --check for safe configuration validation, EXIT CLUSTER for online node removal, dict=keywords_32k for indexing long machine-generated tokens, a built-in Ukrainian lemmatizer, and improved Systemd integration.
  • Extensive Bug Fixes: The release incorporates 65 changelog-listed fixes, addressing critical issues such as crashes with columnar float_vector attributes, replication problems, KNN reordering, and MySQL protocol compatibility.

This consolidated release series for Manticore Search delivers substantial advancements across its core functionalities, making it a more secure, scalable, and intelligent platform for demanding search and data analytics workloads.