AWS

Amazon ECS now integrates directly with Amazon VPC Lattice to manage traffic shifting during service updates. This feature enables blue/green, linear, and canary deployment strategies for applications communicating across VPCs and accounts. Teams can validate new versions with controlled test traffic before fully shifting production load.

• Native traffic shifting for blue/green, linear, and canary strategies within ECS services.
• Supports service-to-service communication across VPCs and AWS accounts via VPC Lattice.
• Enables controlled validation with test traffic before full production rollout.
• Allows custom validation steps or manual approval gates during deployment.

AWS has expanded Amazon DynamoDB Accelerator (DAX) coverage to 17 additional regions, including new locations in Asia Pacific, Europe, Canada, and GovCloud. DAX serves as a fully managed in-memory cache for DynamoDB, capable of delivering up to 10x performance gains for read-heavy workloads. This update extends low-latency, microsecond response capabilities to more global geographies.

• DAX is now live in 17 new regions, expanding global coverage for DynamoDB caching.
• GovCloud (US-East and US-West) now supports DAX for compliant workloads.
• Read-heavy DynamoDB apps can achieve microsecond latency and 10x throughput improvements.
• New regions include key markets like Tokyo, Seoul, Zurich, and Hyderabad.

Aurora PostgreSQL now allows direct querying of operational data alongside lake files in Apache Iceberg and Parquet formats. You can create PostgreSQL foreign tables pointing to S3, S3 Tables, or AWS Glue Data Catalog without ETL pipelines or data duplication. This eliminates the need to copy data into the database for analytics, reducing engineering overhead and storage costs.

• Query Iceberg/Parquet files in S3 directly using standard PostgreSQL foreign tables
• No ETL pipelines required to move lake data into Aurora for analytics
• Supports integration with AWS Glue Data Catalog and S3 Tables
• Reduces data duplication and engineering effort for hybrid workloads

AWS Glue now supports system-managed materialized views for Apache Iceberg, ensuring that precomputed query results remain immutable. This feature restricts write access to the view's data and definition solely to AWS Glue, preventing unauthorized modifications by users with S3 or catalog write permissions. Users define the view via SQL and can optionally schedule automatic refreshes, with results stored as standard Iceberg tables in Amazon S3.

• Prevents accidental or malicious overwrites of precomputed materialized view data.
• Glue is the exclusive writer for the view, enforcing data integrity at the engine level.
• Compatible with existing open Iceberg tables stored in Amazon S3.
• Supports optional refresh schedules for automated data updates.

Amazon Aurora serverless has updated its scaling behavior to add up to 16 ACUs within a single second, enabling it to reach a maximum of 256 ACUs more rapidly. This change is designed to handle the unpredictable, bursty traffic patterns typical of agentic AI applications. The feature is enabled by default on platform versions 3 and 4, requiring no configuration changes.

• Scales up to 16 ACUs per second to handle sudden traffic spikes
• Max capacity extended to 256 ACUs for heavier bursty workloads
• Optimized for agentic AI apps with long idle windows and bursts
• Enabled by default on Aurora platform versions 3 and 4
• Existing clusters on v1/v2 can upgrade to v4 for this capability