Amazon OpenSearch Serverless vector store is the default vector database that Knowledge Bases for Amazon Bedrock provisions when you allow Bedrock to create the store automatically. It uses the OpenSearch Serverless vector engine to store embeddings and perform similarity search for semantic queries.

The OpenSearch Serverless vector store is selected because it provides a managed vector engine that integrates with Knowledge Bases for Bedrock and it scales to large embedding volumes without requiring you to operate a separate vector database. This lets Bedrock handle provisioning, indexing, and similarity search so your team does not need to manage the underlying vector infrastructure.

Amazon DynamoDB is a key value and document database and it is not offered as a native vector store for Knowledge Bases. It lacks the built in vector search engine that OpenSearch Serverless provides and therefore it is not the default.

Amazon Aurora can be configured with pgvector on PostgreSQL as a customer managed integration, but it is not automatically created by Bedrock. That makes Aurora a possible custom option but not the Bedrock managed default.

Redis Enterprise Cloud is available as an external connector for vector storage yet it is not provisioned by Bedrock when the Knowledge Base is created. It remains an external or customer supplied vector service and so it is not the default choice.

Full AWS Practitioner Certification Question