Amazon OpenSearch Service now supports OpenSearch version 3.1

You can now run OpenSearch version 3.1 in Amazon OpenSearch Service. OpenSearch 3.1 introduces several improvements in areas like search relevance and performance, and introduces features that simplify development of vector-driven applications for generative AI workloads.

This launch incorporates Lucene 10 that enables optimized vector field indexing resulting in faster indexing times and reduced index sizes, sparse indexing for CPU and storage efficiency improvements, and vector quantization to reduce memory usage. Other key areas of improvement include improved range query performance, which benefits log analytics and time-series workloads, and reduced latency for high-cardinality aggregations.

This launch also introduces a new Search Relevance Workbench, which provides integrated tools for teams to evaluate and optimize search quality through experimentation. Additionally, this launch includes several improvements in vector search capabilities. First, Z-score normalization improves hybrid search reliability by reducing the impact of outliers and different score scales. Finally, you can now boost efficiency of searches using memory-optimized search that enables the Faiss engine to operate efficiently by memory-mapping the index file and using the operating system's file cache to serve search requests.

For information on upgrading to OpenSearch 3.1, please see the documentation. OpenSearch 3.1 is now available in all AWS Regions where Amazon OpenSearch Service is available.