Posted On: Nov 19, 2024

Amazon OpenSearch Service now supports OpenSearch version 2.17 in Amazon OpenSearch Service. With OpenSearch 2.17, we have made several improvements in the areas of vector search, query performance and machine learning (ML) toolkit to help accelerate application development and enable generative AI workloads.

This launch introduces disk-optimized vector search, a new option for the vector engine that's designed to run efficiently with less memory to deliver accurate, economical vector search at scale. In addition to this, OpenSearch’s FAISS engine now supports byte vectors lowering cost and latency by compressing k-NN indexes with minimal recall degradation. You can now encode numeric terms as a roaring bitmap that enables you to perform aggregations, filtering and more, with lower retrieval latency and reduced memory usage.

This launch also includes key features to help you build ML-powered applications. Firstly, with ML inference search processors you can now run model predictions while executing search queries. In addition to this, you can also perform high-volume ML tasks, such as generating embeddings for large datasets and ingesting them into k-NN indexes using asynchronous batch ingestion. Finally, this launch adds threat intelligence capabilities to the Security Analytics solution. This functionality enables you to use customized Structured Threat Information Expression (STIX)-compliant threat intelligence feeds by uploading a file locally or referencing an Amazon S3 bucket. This functionality offers a more comprehensive view into security threats, affording greater insights to support decision-making and remediation.

OpenSearch 2.17 is now supported on Amazon OpenSearch Service in Amazon Web Services China (Beijing) Region, operated by Sinnet and Amazon Web Services China (Ningxia) Region, operated by NWCD.

For more information about third party plugins, please see the documentation. To learn more about Amazon OpenSearch Service, please visit the product page.