Amazon DynamoDB now supports real-time vector search at any scale

Amazon DynamoDB now supports native vector search in general availability across all commercial AWS Regions. Developers can store vector embeddings alongside operational data in DynamoDB and run similarity searches directly, eliminating the need for a separate vector database or synchronization pipeline. The feature offers single-digit millisecond latency at 99%+ recall, supports up to 4096 dimensions, and works with Euclidean, Cosine, and Dot product distance functions. A new vector index type is created on an attribute storing embeddings, and the SearchVectors API returns ranked results with optional inline filtering. Use cases include RAG, recommendation engines, anomaly detection, and agentic memory. Pricing follows DynamoDB's existing pay-per-request model with no infrastructure to manage.

Nguồn: https://aws.amazon.com/blogs/aws/amazon-dynamodb-now-supports-real-time-vector-search-at-any-scale. 8sync News chỉ tóm tắt và dẫn link; bản quyền nội dung thuộc tác giả và nguồn gốc.