AWS Introduces Native Vector Search for DynamoDB
| Source: InfoQ AI/ML
Tags: DynamoDB, AWS, vector-search, RAG, Amazon Bedrock, embeddings
Amazon DynamoDB now supports native vector search via a new SearchVectors API, letting developers store embeddings and run approximate nearest-neighbor queries in the same table — eliminating the need for a separate vector database for RAG, semantic search, and recommendation workloads.
Details
DynamoDB's new vector search feature uses a dedicated index type built on vector embeddings stored directly in table attributes. Developers choose any embedding model — Amazon Bedrock Titan, Cohere Embed, or OpenAI text embedding models — create a vector index with their preferred dimensions and distance function (Euclidean, Cosine, or Dot product), and query it via the new SearchVectors API. The feature supports inline filtering and up to 4,096 dimensions. Before this, teams building semantic search or RAG applications on DynamoDB had to maintain a separate vector database, copy data into it, and keep the two systems synchronized — a common source of operational friction. Native vector search removes that extra pipeline. Pricing adds three meters on top of standard DynamoDB charges: data written to the index, data processed during searches, and data stored — all metered per byte and billed per GB. AWS notes that lower dimensions, minimal projections, and excluding embeddings from results keep costs down. The feature is fully serverless and scales horizontally without infrastructure management.