As a Cloud Security Architect and DevSecOps Lead, I'm pleased to present today's AWS updates, focusing on enhanced performance, security, and automation capabilities, along with their critical Terraform integration status.
End User Computing
- Amazon WorkSpaces - Graphics G7 Bundles: Amazon WorkSpaces Personal and Core now offer new Graphics G7 bundles, leveraging NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs and Intel Xeon 6 processors. These bundles deliver up to 2.1x performance improvement for graphics-intensive workloads like CAD/CAM, 3D rendering, and AI-assisted design. Available in four sizes, they support Windows with BYOL, AlwaysOn, and AutoStop modes in select regions. From a security and architectural perspective, these high-performance instances demand robust data-at-rest encryption (via KMS) for sensitive visual data and careful selection of bundle sizes to meet performance and cost requirements. Compliance with data residency is also a key consideration for region selection. *Terraform Support*: Supported in Terraform AWS Provider:
aws_workspaces_workspacein v5.68.0+.
Networking
- AWS Client VPN - MacOS 27 Golden Gate Client Support: AWS Client VPN now supports the MacOS 27 Golden Gate client (versions 6.0+), expanding compatibility for the AWS-supplied VPN client to the latest macOS versions. This ensures that users on cutting-edge Apple hardware and software can maintain secure remote access. Architecturally, this update simplifies client management for organizations adopting the latest macOS, ensuring a consistent and secure user experience. Maintaining up-to-date client software is a fundamental security practice. *Terraform Support*: Supported in Terraform AWS Provider:
aws_ec2_client_vpn_endpointin v5.76.0+.
Contact Center
- Amazon Connect - Tag-Based Access Control for Custom Metrics: Amazon Connect custom metrics now feature tag-based access control, enabling granular governance over which users and roles can view, create, or modify specific custom metrics. This is a significant security enhancement, allowing organizations to implement Attribute-Based Access Control (ABAC) for operational insights. It facilitates multi-team collaboration while enforcing strict separation of duties and compliance with internal data governance policies. A well-defined tagging strategy is crucial for effective implementation. *Terraform Support*: Supported for defining IAM policies with tag conditions in Terraform AWS Provider:
aws_iam_policyin v5.61.0+. - Amazon Connect - Custom Metrics APIs: Amazon Connect has introduced seven new API operations (CreateMetric, DeleteMetric, DescribeMetric, ListMetrics, SearchMetrics, UpdateMetricContent, and UpdateMetricMetadata) for programmatic management of custom metrics. This allows for consistent metric definitions across environments and enables automation for monitoring dashboards. All changes are meticulously tracked in AWS CloudTrail, providing a comprehensive audit trail for security and compliance. Architecturally, these APIs facilitate integration into CI/CD pipelines for metric lifecycle management. It is critical to apply least-privilege IAM policies to any entity interacting with these new APIs. *Terraform Support*: Pending provider release (use AWS CLI/SDK or custom provisioner in the interim).
Machine Learning
- Amazon SageMaker AI - Serverless Customization for NVIDIA Nemotron 3.5 Lightning: Amazon SageMaker AI now supports serverless model customization for the NVIDIA Nemotron 3.5 Lightning model using supervised fine-tuning (SFT), Direct Preference Optimization (DPO), and reinforcement fine-tuning (RFT). This serverless approach abstracts infrastructure management, allowing users to pay only for actual usage. Security considerations include ensuring robust IAM roles for SageMaker, encrypting training data and model artifacts with KMS, and isolating training jobs within private networks. This streamlines the secure deployment of advanced AI models while adhering to data handling regulations. *Terraform Support*: Supported for deploying the customized model in Terraform AWS Provider:
aws_sagemaker_modelin v5.60.0+.
Security, Identity, & Compliance
- AWS Security Token Service (STS) - Unified 4,096-byte Session Token Limit: AWS STS now enforces a unified 4,096-byte size limit for session tokens, providing significantly more capacity for rich session configurations, including inline policies, managed policies, and session tags. STS API responses now include token size and utilization, automatically logged in AWS CloudTrail and published as metrics in Amazon CloudWatch. An optional API parameter is available for testing larger token sizes. This enhancement is a critical security improvement, enabling more granular and robust Attribute-Based Access Control (ABAC) implementations. The detailed logging and monitoring capabilities in CloudTrail and CloudWatch are invaluable for auditing, detecting anomalous activity, and ensuring compliance. Architects should leverage this increased capacity to implement more precise least-privilege temporary credentials. *Terraform Support*: Supported for managing underlying IAM roles and policies that influence token size in Terraform AWS Provider:
aws_iam_rolein v5.25.0+.
Containers
- Amazon ECS - S3 Files Support for EC2 Launch Type: Amazon ECS now supports Amazon S3 Files for tasks running on the Amazon EC2 launch type, enabling containerized applications to access data in Amazon S3 as a shared file system with standard file system semantics. This eliminates the need for code changes or data staging, providing consistent S3 data access across all ECS launch types (Fargate, Managed Instances, and now EC2). Built on Amazon EFS, S3 Files offers low-latency performance without data leaving S3. Security considerations include configuring least-privilege IAM roles for ECS tasks to access S3 and EFS, enforcing KMS encryption for S3 buckets and EFS file systems, and applying strict network access controls (Security Groups). This simplifies data management for containerized workloads while maintaining strong security postures. *Terraform Support*: Supported for defining volume mounts in Terraform AWS Provider:
aws_ecs_task_definitionin v5.65.0+.
Media Services
- AWS Elemental MediaTailor - New Lifecycle Hooks for Monetization: AWS Elemental MediaTailor now offers two new lifecycle hooks:
post ad decision server (ADS) responseandpre manifest insertion. These hooks allow customers to programmatically intervene in the ad insertion workflow, enabling advanced monetization strategies such as calling secondary ad sources, filtering ads based on content/brand policies, adding house ads, or performing final checks before ad pod delivery. The fail-open design ensures viewer playback remains unaffected by hook errors. From a security perspective, any Lambda functions or external services invoked by these hooks must be secured with least-privilege IAM roles, robust logging, and input validation. These hooks provide powerful tools for compliance with advertising standards and content policies. *Terraform Support*: Supported for configuring playback configurations with hooks in Terraform AWS Provider:aws_mediatailor_playback_configurationin v5.40.0+.