AWS prescriptive-guidance documentation change
Summary
Major restructuring and expansion of vector database options documentation, adding detailed sections for Amazon DocumentDB, Amazon MemoryDB, Amazon Neptune Analytics, and Amazon S3 Vectors, while reorganizing the managed service option section for Amazon Bedrock Knowledge Bases and adding a new 'Choosing the right vector database' guidance section.
Security assessment
This is a comprehensive documentation update that expands service coverage and provides detailed guidance on selecting vector databases. There is no evidence of addressing a specific security vulnerability, weakness, or incident. The security-related content (IAM controls, encryption, VPC support, CloudTrail logging) was already present in the original document and remains unchanged in the updated managed service section. The changes are focused on feature descriptions, performance characteristics, and selection criteria rather than security fixes or new security features.
Diff
diff --git a/prescriptive-guidance/latest/choosing-an-aws-vector-database-for-rag-use-cases/vector-db-options.md b/prescriptive-guidance/latest/choosing-an-aws-vector-database-for-rag-use-cases/vector-db-options.md index 841771de4..8c67ce6c3 100644 --- a//prescriptive-guidance/latest/choosing-an-aws-vector-database-for-rag-use-cases/vector-db-options.md +++ b//prescriptive-guidance/latest/choosing-an-aws-vector-database-for-rag-use-cases/vector-db-options.md @@ -5 +5 @@ -Individual vector database optionsManaged service option +Individual vector database optionsManaged service optionChoosing the right vector database @@ -16,0 +17,2 @@ For more information about vector database solutions, see the following sections + * Choosing the right vector database + @@ -22 +24 @@ For more information about vector database solutions, see the following sections -The individual vector database options on AWS include Amazon Kendra, Amazon OpenSearch Service, and Amazon RDS for PostgreSQL with pgvector. (An open-source extension, pgvector adds the ability to store and search machine learning (ML)-generated vector embeddings.) These solutions offer different approaches to vector search, allowing organizations to choose based on their existing infrastructure, technical requirements, and specific [use cases](./use-cases.html). +The individual vector database options on AWS include [Amazon Kendra](https://docs.aws.amazon.com/kendra/latest/dg/what-is-kendra.html), [Amazon OpenSearch Service](https://docs.aws.amazon.com/opensearch-service/latest/developerguide/what-is.html), [Amazon RDS for PostgreSQL](https://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/CHAP_PostgreSQL.html) with pgvector, [Amazon MemoryDB](https://docs.aws.amazon.com/memorydb/latest/devguide/what-is-memorydb.html), [Amazon DocumentDB](https://docs.aws.amazon.com/documentdb/latest/developerguide/what-is.html), [Amazon Neptune Analytics](https://docs.aws.amazon.com/neptune-analytics/latest/userguide/what-is-neptune-analytics.html), and [Amazon S3 Vector](https://docs.aws.amazon.com/AmazonS3/latest/userguide/s3-vectors.html). (An open-source extension, pgvector adds the ability to store and search ML-generated vector embeddings.) These solutions offer different approaches to vector search, allowing organizations to choose based on their existing infrastructure, technical requirements, and specific [use cases](./use-cases.html). @@ -56 +58 @@ Benefits of Amazon Kendra include the following -For more information, see [Benefits of Amazon Kendra](https://docs.aws.amazon.com/kendra/latest/dg/what-is-kendra.html#what-is-benefits) in the _Amazon Kendra Developer Guide_. +For more information, see [Benefits of Amazon Kendra](https://docs.aws.amazon.com/kendra/latest/dg/what-is-kendra.html#what-is-benefits) in the Amazon Kendra documentation. @@ -88 +90 @@ Some advantages of using OpenSearch Service include the following: -For more information, see [Features of Amazon OpenSearch Service](https://docs.aws.amazon.com/opensearch-service/latest/developerguide/what-is.html#what-is-features) in the _OpenSearch Service Developer Guide_. +For more information, see [Features of Amazon OpenSearch Service](https://docs.aws.amazon.com/opensearch-service/latest/developerguide/what-is.html#what-is-features) in the OpenSearch Service documentation. @@ -109 +111 @@ Key benefits of Amazon RDS for PostgreSQL with pgvector include the following: -For more information, see [Advantages of Amazon RDS](https://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/Welcome.html) in the _Amazon Relational Database Service User Guide_. +For more information, see [Advantages of Amazon RDS](https://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/Welcome.html) in the Amazon RDS documentation. @@ -111 +113 @@ For more information, see [Advantages of Amazon RDS](https://docs.aws.amazon.com -## Managed service option +### Amazon DocumentDB @@ -113 +115 @@ For more information, see [Advantages of Amazon RDS](https://docs.aws.amazon.com -Amazon Bedrock Knowledge Bases represents the AWS fully managed approach to vector database implementation. The service's flexibility in storage options, combined with its automated management features, makes it particularly valuable for organizations seeking to implement RAG without managing complex infrastructure. +Amazon DocumentDB (with MongoDB compatibility) is a document database that offers native vector search capabilities in version 5.0 and later. It combines the flexibility of JSON-based document storage with vector search, supporting both hierarchical navigable small world (HNSW) and Inverted File Flat (IVFFlat) indexing methods. @@ -115 +117 @@ Amazon Bedrock Knowledge Bases represents the AWS fully managed approach to vect -With Amazon Bedrock Knowledge Bases, you can create, maintain, and query knowledge bases that enhance your foundation models using RAG. This service simplifies the complex process of implementing RAG by managing the entire data ingestion, vectorization, and retrieval pipeline. +Core capabilities of Amazon DocumentDB include the following: @@ -117 +119 @@ With Amazon Bedrock Knowledge Bases, you can create, maintain, and query knowled -Key benefits of Amazon Bedrock Knowledge Bases include the following: + * Store and index vectors up to 2,000 dimensions (up to 16,000 dimensions without indexing) @@ -119 +121 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: -**Simplified data processing** + * Millisecond response times for vector similarity searches @@ -121 +123 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: - * Automatic data ingestion and chunking + * Support for euclidean, cosine, and dot product distance metrics @@ -123 +125 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: - * Built-in text extraction from multiple file formats + * Seamless integration with existing MongoDB-compatible applications @@ -125 +126,0 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: - * Managed vector embeddings generation @@ -127 +127,0 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: - * Automatic metadata extraction and indexing @@ -129,0 +130 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: +Use Amazon DocumentDB in the following situations: @@ -130,0 +132 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: + * For applications that are already using MongoDB APIs and that need vector search capabilities @@ -132 +134 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: -**Streamlined RAG implementation** + * For use cases that require flexible document data structures combined with semantic search @@ -134 +136 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: - * Pre-configured retrieval strategies + * For scenarios that need both traditional document queries and vector similarity searches @@ -136 +138 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: - * Automatic context window optimization + * For applications that provide product recommendations, personalization, chat assistants, and fraud detection @@ -138 +139,0 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: - * Built-in relevancy tuning @@ -140 +140,0 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: - * Semantic search capabilities out of the box @@ -142,0 +143 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: +For more information, see [Vector search for Amazon DocumentDB](https://docs.aws.amazon.com/documentdb/latest/developerguide/vector-search.html) in the Amazon DocumentDB documentation. @@ -143,0 +145 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: +### Amazon MemoryDB @@ -145 +147 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: -**Security and governance** +Amazon MemoryDB is a Redis-compatible, in-memory database that delivers the fastest vector search performance among popular vector databases on AWS. It provides sub-millisecond query latencies with multi-Availability Zone durability. @@ -147 +149 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: - * Integrated AWS Identity and Access Management (IAM) controls +Core capabilities of MemoryDB include the following: @@ -149 +151 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: - * Data encryption at rest and in transit + * Store application data and millions of vectors in a single database @@ -151 +153 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: - * VPC support + * Single-digit millisecond query and update response times @@ -153 +155 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: - * Audit logging with AWS CloudTrail + * Highest recall rates at the fastest performance on AWS @@ -154,0 +157 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: + * Support for up to 32,768 dimensions per vector @@ -155,0 +159 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: + * Real-time semantic search and caching capabilities @@ -158 +161,0 @@ Key benefits of Amazon Bedrock Knowledge Bases include the following: -Amazon Bedrock Knowledge Bases supports multiple [vector store options](https://docs.aws.amazon.com/bedrock/latest/userguide/knowledge-base-setup.html). The following list provides an overview of each option's key features: @@ -160 +162,0 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Amazon Aurora PostgreSQL with pgvector @@ -162 +164 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * PostgreSQL-compatible vector storage +Use MemoryDB in the following situations: @@ -164 +166 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Integrated with existing Aurora databases + * For real-time applications that require ultra-low latency (sub-10ms) @@ -166 +168 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Cost-effective for smaller deployments + * For high-throughput workloads with millions of requests per day @@ -168 +170 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Good for hybrid structured and unstructured data + * For use cases such as real-time recommendation engines, semantic caching, and anomaly detection @@ -170 +172 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Amazon Neptune Analytics + * For applications that need both in-memory data store and vector search capabilities @@ -172 +173,0 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Graph-based vector search @@ -174 +174,0 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Combines relationship data with vectors @@ -176 +175,0 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Ideal for connected data use cases @@ -178 +177 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Advanced query capabilities +For more information, see [Vector search](https://docs.aws.amazon.com/memorydb/latest/devguide/vector-search.html) in the MemoryDB documentation. @@ -180 +179 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Amazon OpenSearch Serverless +### Amazon Neptune Analytics @@ -182 +181 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Fully managed serverless experience +Amazon Neptune Analytics is a graph analytics engine that offers native vector search capabilities, making it ideal for Graph Retrieval Augmented Generation (GraphRAG) use cases. It combines vector similarity search with graph traversals and algorithms. @@ -184 +183 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Automatic scaling based on workload +Core capabilities of Neptune Analytics include the following: @@ -186 +185 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Built-in k-NN capabilities + * Analyze tens of billions of relationships within seconds @@ -188 +187 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Cost-effective for varying workloads + * Combine vector search with graph algorithms (path finding, community detection, centrality) @@ -190 +189 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Pinecone + * Support for GraphRAG applications with topological knowledge @@ -192 +191 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Purpose-built vector database + * Up to 80 times faster than existing graph analytical solutions @@ -194 +193 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * High performance at scale + * Integration with Amazon Bedrock for fully managed GraphRAG @@ -196 +194,0 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Advanced similarity search features @@ -198 +195,0 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Managed through the Amazon Bedrock console @@ -200 +196,0 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Redis Enterprise Cloud @@ -202 +198,31 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * In-memory vector search capabilities +Use Neptune Analytics in the following situations: + + * For GraphRAG applications that require knowledge graphs with vector embeddings + + * For use cases that require traversing complex relationships alongside vector similarity + + * For applications that require explainable AI responses with relationship context + + * For scenarios such as customer 360 views, fraud detection networks, and knowledge discovery + + + + +For more information, see the [Amazon Neptune Analytics documentation](https://docs.aws.amazon.com/neptune-analytics/latest/userguide/what-is-neptune-analytics.html). + +### Amazon S3 Vectors + +Amazon S3 Vectors is the first cloud object store in AWS with native vector storage and query capabilities. It provides purpose-built, cost-optimized vector storage for AI applications that require massive scale. + +Core capabilities of Amazon S3 Vectors include the following: + + * Storage for up to 2 billion vectors per index with support for up to 10,000 indexes per vector bucket + + * Sub-100 ms query latency that is optimized for long-term storage and infrequent access patterns + + * Up to 90% cost reduction for vector operations compared to specialized vector databases + + * Serverless architecture with automatic scaling and 99.999999999% (11 9s) durability + + + @@ -204 +230 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Low-latency performance +Use Amazon S3 Vectors in the following situations: @@ -206 +232 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Real-time vector search + * For applications that require storage of billions of vectors at minimal cost @@ -208 +234 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// - * Integrated caching capabilities + * For workloads that tolerate sub-second query latency (100 ms or more) rather than sub-10 ms @@ -209,0 +236 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// + * For long-term vector retention and archival use cases @@ -210,0 +238 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// + * For RAG applications with infrequent retrieval patterns @@ -211,0 +240 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// + * For organizations that prioritize storage economics over ultra-low latency @@ -213 +241,0 @@ Amazon Bedrock Knowledge Bases supports multiple [vector store options](https:// -When choosing a vector store that's supported by Amazon Bedrock Knowledge Bases, consider the following key characteristics of each option: @@ -215 +242,0 @@ When choosing a vector store that's supported by Amazon Bedrock Knowledge Bases, - * **Aurora PostgreSQL** – Relational data with vector capabilities @@ -217 +243,0 @@ When choosing a vector store that's supported by Amazon Bedrock Knowledge Bases, - * **Neptune Analytics** – Graph-based knowledge representations @@ -219 +245 @@ When choosing a vector store that's supported by Amazon Bedrock Knowledge Bases,