AWS

Mountpoint for Amazon S3 now supports explicit memory usage controls, allowing administrators to define hard limits or rely on automatic environment-based defaults. This update addresses previous issues where memory consumption could grow unbounded, potentially destabilizing co-located memory-intensive workloads like ML training. By reserving memory for primary applications, systems running Mountpoint in shared environments such as Amazon EKS can maintain stability and predictable performance.

• Define explicit memory targets for Mountpoint to prevent resource contention.
• Enable automatic memory detection for safe defaults in environments like EKS.
• Stabilize workloads where Mountpoint shares memory budgets with ML or analytics apps.
• Mitigate performance degradation caused by unbounded memory growth in Mountpoint.

Amazon EC2 R8id instances are now live in Asia Pacific, Canada, and Europe. These instances deliver up to 22.8TB of NVMe local storage and 3.3x higher memory bandwidth compared to the R6id generation. They are optimized for memory-intensive workloads such as in-memory databases and real-time analytics. Users can also adjust network and EBS bandwidth by 25% via instance bandwidth weighting.

• R8id instances are now available in Mumbai, Malaysia, Sydney, Canada Central, Ireland, and Stockholm.
• Local NVMe storage capacity reaches 22.8TB, triple that of the previous R6id generation.
• Memory bandwidth improves by 3.3x, offering up to 43% higher performance for memory-heavy tasks.
• Instance bandwidth weighting allows 25% adjustments to network and EBS throughput configurations.

AWS Glue now supports catalog federation for remote Apache Iceberg REST catalogs within AWS GovCloud US regions. This feature enables direct, secure access to Iceberg tables stored in S3 and managed by external catalogs without requiring data movement. Metadata synchronization occurs in real-time between the AWS Glue Data Catalog and the remote source, ensuring query results reflect the latest table definitions.

• Eliminates data duplication by federating directly to remote Iceberg REST catalogs in GovCloud.
• Maintains real-time metadata sync so queries always see the latest table schema and partition info.
• Preserves existing security configurations while allowing analytics engines to access remote data.
• Enables price-performance optimization by leveraging preferred AWS analytics engines on remote data.

AWS has introduced a specification-driven approach for composing data workflows, separating business intent from the underlying processing logic. The architecture relies on declarative specifications, reusable processing capabilities, and pre-execution validation to ensure compliance. This method aims to cut dataset onboarding time from weeks to days while providing robust traceability, versioning, and governance features.

• Declarative specs separate intent from logic, enabling faster, more flexible data workflow composition.
• Pre-execution validation and reusable capabilities streamline governance and reduce onboarding time.
• Built-in traceability and versioning support stricter compliance requirements for data pipelines.

AWS Backup now enables cross-Region backup copying and logically air-gapped vault storage for Amazon DocumentDB across nine new regions. This update allows users to implement robust disaster recovery and compliance strategies by isolating immutable backups. The feature supports both automated backup plans and on-demand copy jobs, with vault sharing available via AWS Resource Access Manager. Encryption is enforced using either AWS-owned or customer-managed keys for added security.

• Cross-Region copy for DocumentDB is now live in AP-HK, AP-JK, AP-MEL, AP-OSA, EU-ESP, EU-STO, EU-ZRH, AF-CPT, and IL-TLV.
• Logically air-gapped vaults provide immutable, isolated storage to protect against ransomware and accidental deletion.
• Use AWS Backup plans or on-demand jobs to manage cross-Region replication for DocumentDB instances.
• Share air-gapped vaults across accounts using AWS Resource Access Manager for streamlined recovery.
• Backups are encrypted by default, supporting both AWS-owned keys and customer-managed CMKs.

AWS has extended AWS PrivateLink support to Lambda MicroVMs, allowing private connectivity from VPC resources without traversing the public internet. This update enables regulated workloads in finance, healthcare, and government to meet strict network isolation requirements. Developers can now create PrivateLink VPC Endpoints to securely call MicroVM APIs directly from their VPCs.

• Enables private VPC connectivity to Lambda MicroVMs without public internet exposure.
• Addresses strict network isolation needs for regulated industries like finance and healthcare.
• Allows creation of PrivateLink endpoints to securely invoke MicroVM APIs from VPCs.

Amazon RDS for Oracle now supports the July 2026 Release Update for versions 19c, 21c, and 26ai, including critical security patches. The naming convention for 19c RUs has shifted to a quarterly format, while monthly Critical Security Patch Updates are bundled into monthly Release Patches. Oracle Database 21c RUs continue to use their existing naming scheme without these changes.

• Upgrade to July 2026 RU for mandatory security fixes across 19c, 21c, and 26ai.
• New 19c RU naming follows <version>.ru-YYYY-MM.mrp-YYYY-MM.rN format.
• Monthly CSPUs for 19c and 26ai are now delivered via monthly Release Patches.
• 21c RUs retain their legacy naming convention for backward compatibility.

Fleet impact: For Oracle ExaCC/RAC and RDS fleets, prioritize patching 19c instances to the July 2026 RU to address security vulnerabilities. Verify your deployment scripts for the new 19c version string format, as existing automation relying on the old naming convention may fail. For 21c environments, continue using standard RU update procedures as the naming structure remains unchanged.