You can now run OpenSearch version 2.19 in Amazon OpenSearch Service which introduces several improvements in the areas of Vector Search, Observability and OpenSearch Dashboards.
We have introduced four key capabilities for vector search applications. The Faiss engine now supports AVX512 SIMD instructions, to accelerate vector similarity computations. The ML inference search response processor can now rank search hits and update scores based on model predictions, enabling sophisticated and context-aware document ranking and result augmentation. Lucene binary vectors , now complement existing Faiss engine binary vector support offering greater flexibility for vector search applications. Hybrid search now includes pagination support , reciprocal rank fusion to improve result ranking along with a debugging tool for score and rank normalization process.
The launch also introduces query insights dashboards that lets users monitor and analyze the top queries collected by the query Insights plugin. Anomaly detection now offers two key improvements. First, enhanced anomaly definition capabilities allow users to specify multiple criteria to identify both spikes and dips in data patterns. Second, a new dedicated index for flattened results improves query performance and dashboard visualization experience. Finally, you can now use template query to create search queries that contain placeholder variables allowing for more flexible, efficient, and secure search operations.
For information on upgrading to OpenSearch 2.19, please see the documentation
. OpenSearch 2.19 is now available in all AWS Regions where Amazon OpenSearch Service is available.