Built on top of Apache Lucene, OpenSearch is an open source, distributed search and analytics engine designed for real-time monitoring, analytics logging, and website search.
It's highly scalable nature allows users to handle large volumes of data across distributed systems, providing fast and powerful search capabilities, making it ideal for a wide range of applications, including AI and machine learning.
On that note, a new major release has arrived after three years, marking a significant evolution of the platform with major improvements across the board.
OpenSearch 3.0: What's New?
A live demo of OpenSearch Dashboards visualizing sample flight data, featuring an alpha version of OpenSearch 3.0.
Taking advantage of Lucene 10, OpenSearch 3.0 delivers major improvements in search and index performance, alongside the introduction of sparse indexing, which allows OpenSearch to efficiently group similar documents together, resulting in better CPU and storage utilization.
For search, this release improves hybrid search by using Z-score normalization and setting a minimum score limit to make results more accurate. It also adds inner hits support to show detailed scores and speeds up aggregations with better star-tree indexing for faster query responses.
As for its vector database capabilities, OpenSearch 3.0 features substantial advancements, with the notable ones being the experimental integration of GPU acceleration using NVIDIA cuVS for powering vector searches, the introduction of native Model Context Protocol (MCP) support, and semantic sentence highlighting.
OpenSearch 3.0 also improves query monitoring with a new Live Queries API for real-time tracking of running queries, a verbose parameter for optimizing dashboard performance, and dynamic columns that adapt to filters. All of these together enable efficient performance analysis and easier identification of resource-heavy queries.
We wrap this up with the deprecation of support for Ubuntu 20.04, which is set to reach the end of its standard support lifecycle on May 31, 2025. The official announcement blog and the release notes have additional information if you are up for a longer read.
Get OpenSearch 3.0
If you liked what you read, then you can try OpenSearch 3.0 with Docker Compose or download the relevant package for your operating system from the official website.
There is support for Linux, FreeBSD, Docker, and Windows, and the documentation is a valuable resource if you need help configuring OpenSearch.
From It's FOSS News via this RSS feed