AWS Database Blog

Organizations running Amazon RDS and Amazon Aurora often pay full CloudWatch Logs ingestion rates for database logs they rarely access. This post shows how to build an automated, tag-driven solution that migrates RDS and Aurora CloudWatch log groups from the Standard to the Infrequent Access log class and cuts log ingestion costs by about 50%.

In this post, you learn how to use Amazon OpenSearch Service as a discovery layer that combines full-text, vector, and geospatial search across JSON documents in a single query, using a restaurant discovery app as a working example.

Recover from accidental changes to your Amazon DynamoDB tables without a full table restore. This post shows how to use the Bulk Executor revert-export command with an incremental export to Amazon S3 to undo unwanted writes, target only a subset of changes with a transform, or fix specific items along the way.

Learn how to configure connection pooling for the AWS Advanced JDBC Wrapper on Amazon Aurora and Amazon RDS. This post explains how the wrapper’s external and internal pooling differ, how to choose between them, and how the JDBC-WRAPPER-CONFIGURATION-ASSISTANT helps you build the right configuration.

Learn how to build zero-downtime write architectures for Amazon Neptune using a write queue pattern with Amazon SQS, Amazon Kinesis Data Streams, or Amazon MSK. By decoupling write acceptance from write execution, your application keeps accepting graph writes during maintenance windows, failovers, and scaling operations.

Learn how to set up the Amazon Aurora DSQL MCP server and use it from your AI coding assistant to run queries, evolve schemas, and check Aurora DSQL compatibility without leaving your IDE. This post walks through installation, the available MCP tools, practical integration patterns, and the security model.

Learn how to build, train, and score a credit risk machine learning model entirely in SQL using Oracle Machine Learning for SQL (OML4SQL) on Amazon RDS for Oracle. The built-in AutoML feature selects the best algorithm automatically, with no data movement and no external machine learning platforms.

AWS DMS uses exponential backoff for recoverable errors, and default settings can let a change data capture (CDC) task retry silently for up to 30 minutes before failing. This post shows how to tune four recoverable-error settings so CDC tasks fail within minutes, and how to pair them with Amazon EventBridge and Amazon CloudWatch alerts.

Many applications that use Amazon DynamoDB for operational data also need vector similarity search, which until now meant running a separate vector database. DynamoDB now supports native vector search. In this post, we build a semantic search application over research paper abstracts using Amazon Bedrock embeddings, and cover how vector search is metered.

Amazon Keyspaces supports two approaches for modeling complex data: user-defined types (UDTs) and Protocol Buffers. This post shows how to create and manage UDTs, implement Protobuf serialization, and choose between the two based on your application needs.