As your Cloud Security Architect, DevSecOps Lead, and Terraform Specialist, here's a comprehensive summary of today's significant AWS feature releases and service updates, focusing on their security implications, architectural considerations, and current Terraform availability.
Generative AI & Machine Learning Highlights
- Amazon Bedrock AgentCore Memory – Flexible Namespace Variables: This new feature allows developers to define flexible namespace variables to scope long-term memories along application-specific dimensions (e.g., organization, tenant, team, environment). This provides fine-grained control over memory organization, isolation, and access, crucial for multi-tenant and complex-hierarchy applications. From a security and architectural perspective, this significantly enhances data isolation and reduces the complexity of managing distinct memory contexts, aiding compliance efforts by ensuring clearer data segregation. *Terraform Support*: Supported in Terraform AWS Provider: aws_bedrock_agentcore_memory (likely via a new argument for
namespace_variablesormemory_keys) in v5.68.0+.
- Amazon Bedrock AgentCore Memory – Fine-Grained Access Control (FGAC): AgentCore Memory now supports fine-grained access control, enabling per-user and per-tenant memory isolation via AgentCore Gateway. By configuring the Gateway for OAuth (JWT) authentication and attaching Cedar policies, access can be restricted based on the authenticated caller's identity and token claims. This is a critical security update, shifting access control enforcement from application code to the infrastructure layer using cryptographic proof of identity. It significantly reduces the attack surface, enhances compliance (e.g., GDPR, HIPAA) by ensuring strict data isolation, and is an architectural best practice for multi-tenant AI applications. *Terraform Support*: Supported in Terraform AWS Provider: aws_bedrock_agentcore_gateway (for configuring the gateway and its authentication/authorization mechanisms) in v5.68.0+.
- Amazon Bedrock – SpaceXAI Grok 4.6 in AWS GovCloud (US): SpaceXAI Grok 4.6, a new large language model, is now available in AWS GovCloud (US). It features a substantial 500k context window and configurable reasoning efforts (low, medium, high, xhigh). This model is accessible via the
bedrock-runtimeendpoint for core Bedrock APIs and supports cross-Region inference routing. Its availability in GovCloud is key for compliance (FedRAMP, DoD SRG) for government and highly regulated workloads. The large context window and configurable reasoning allow for more complex, sensitive tasks while maintaining control over resource usage and output quality, and cross-Region inference enhances resilience. *Terraform Support*: Supported for related configurations like logging via Terraform AWS Provider: aws_bedrock_model_invocation_logging_configuration in v4.59.0+.
- Amazon SageMaker – NVIDIA Cosmos3 Models on JumpStart: NVIDIA's Cosmos3-Edge, Cosmos3-Nano, and Cosmos3-Super models are now available on Amazon SageMaker JumpStart. These omnimodal world models are designed for physical AI, enabling robots, autonomous vehicles, and vision AI to perceive, reason, plan, and act in the physical world. Deploying these models via SageMaker JumpStart leverages SageMaker's inherent security features, including VPC isolation, KMS encryption for data at rest and in transit, and IAM for access control. For physical AI, ensuring the integrity and security of model outputs is paramount, especially in safety-critical applications. *Terraform Support*: Supported for endpoint deployment via Terraform AWS Provider: aws_sagemaker_endpoint in v4.67.0+.
- Amazon SageMaker – Muse-Glimmer-30B and Qwen 3.8-27B on JumpStart: Two new foundation models, Muse-Glimmer-30B (Meta Superintelligence Lab) and Qwen 3.8-27B (Alibaba), are now available on SageMaker JumpStart. Muse-Glimmer-30B is engineered for autonomous agentic tasks with multi-step reasoning and tool use, featuring a 131K+ context window. Qwen 3.8-27B is a native vision-language model excelling in coding, agentic tasks, and multimodal understanding with a 262K context window. Similar to other JumpStart models, SageMaker's robust security posture applies. For agentic tasks, robust input validation and output sanitization are crucial to prevent prompt injection or unintended actions. Muse-Glimmer's ability to operate without cloud infrastructure introduces unique edge security considerations, requiring secure deployment and update mechanisms. *Terraform Support*: Supported for endpoint configuration via Terraform AWS Provider: aws_sagemaker_endpoint_configuration in v2.0.0+.
Compute Updates
- Amazon EC2 – P6-B300 Instances in AP (Hyderabad) and SA (Sao Paulo): Amazon EC2 P6-B300 instances are now available in Asia Pacific (Hyderabad) and South America (Sao Paulo) Regions. These instances offer 8x NVIDIA Blackwell Ultra GPUs, 2.1 TB HBM, 6.4 Tbps EFA networking, and 4 TB of system memory, providing significant performance gains for training and deploying large foundation models. This regional expansion improves data residency options and reduces latency for local users. The high-bandwidth networking (EFA, ENA) is critical for distributed training, requiring careful network security group and NACL configurations. Data at rest (EBS) and in transit should always be encrypted. *Terraform Support*: Supported for instance provisioning via Terraform AWS Provider: aws_instance (using
instance_type = "p6.b300.24xlarge"or similar) in v5.60.0+.
- Amazon EC2 – C8gn Instances in Europe (Paris): Amazon EC2 C8gn instances, powered by AWS Graviton4 processors, are now available in the AWS Europe (Paris) region. These instances offer up to 30% better compute performance than Graviton3-based C7gn instances, up to 600 Gbps network bandwidth, and support for Elastic Fabric Adapter (EFA). This regional expansion supports data residency and latency requirements for European customers. Graviton4 offers potential cost savings and improved energy efficiency. The high network bandwidth necessitates robust network security configurations, and EFA for HPC/ML workloads demands careful network design. *Terraform Support*: Supported for instance provisioning via Terraform AWS Provider: aws_instance (using
instance_type = "c8gn.xlarge"or similar) in v5.10.0+.
Database Enhancements
- Amazon Aurora MySQL 3.13 (MySQL 8.0.45 Compatible): Amazon Aurora MySQL 3.13, compatible with MySQL 8.0.45, is now generally available. This release includes critical community MySQL fixes and Aurora improvements. A significant operational enhancement is the ability to leverage automatic minor version upgrades and the AWS Organizations Upgrade Rollout Policy for orchestrated, phased upgrades across multiple Aurora clusters. Staying current with MySQL community fixes is crucial for security patching. The Upgrade Rollout Policy is a significant operational security benefit, allowing controlled, automated patching across an organization, reducing human error and downtime risk. Encryption at rest (KMS) and in transit should always be enforced. *Terraform Support*: Supported for cluster configuration via Terraform AWS Provider: aws_rds_cluster (setting
engine_version = "8.0.45"or3.13) in v4.0.0+.
- Amazon Aurora DSQL – Foreign Key Constraints: Aurora DSQL clusters now support adding foreign key constraints to both new and existing tables. This enables developers to express relationships between tables, with the database automatically preventing writes that would violate these constraints. Users can specify actions (NO ACTION, RESTRICT, CASCADE, SET NULL, SET DEFAULT) for referenced row deletions or updates. This is a critical data integrity feature, fundamental for maintaining data consistency and preventing logical corruption. It significantly improves data quality for applications requiring strong relational integrity, reducing application-level validation complexity and potential for data errors. *Terraform Support*: Pending provider release (foreign key constraints are typically managed via DDL within the database, not directly by the
aws_rds_clusterresource, which manages the infrastructure). Use AWS CLI/SDK or custom provisioner in the interim for DDL execution. Terraform AWS Provider: aws_rds_cluster.
Monitoring & Observability
- Amazon CloudWatch Agent – Systemd Journal (journald) Logs: AWS now supports collecting systemd journal (journald) logs directly with the Amazon CloudWatch agent. This allows configuring the agent to read log entries directly from the systemd journal on Linux instances and send them to CloudWatch Logs, preserving structured metadata. This is an improved logging integrity feature, as direct ingestion from journald reduces the risk of log tampering or loss that can occur when logs are written to disk first. Preserving structured metadata enhances forensic analysis and security monitoring. Centralized logging in CloudWatch Logs with KMS encryption is a security best practice. *Terraform Support*: Supported for target log group management via Terraform AWS Provider: aws_cloudwatch_log_group in v5.20.0+ (CloudWatch Agent configuration itself is typically managed via User Data or a configuration management tool).
Migration and Modernization
- AWS Transform – FedRAMP Class C Scope: AWS Transform is now in scope for FedRAMP Class C (formerly Moderate baseline) in the US East (Ohio) Region. This allows customers to use AWS Transform for applications and workloads subject to FedRAMP Class C compliance requirements. This is a critical compliance update, significantly expanding the utility of AWS Transform for government agencies and contractors. Architects must ensure that workloads built with Transform also adhere to FedRAMP controls, with data encryption, access control, and audit logging remaining paramount. *Terraform Support*: Terraform manages the underlying infrastructure for AWS Transform; the service's compliance scope is not a direct Terraform resource attribute. Terraform AWS Provider Documentation in v5.48.0+.
Contact Center Innovations
- Amazon Connect – Automatic Scheduling Metrics Refresh: Amazon Connect now automatically refreshes scheduling metrics on the scheduling page, providing managers immediate visibility into the impact of schedule updates. While primarily an operational efficiency feature, timely and accurate operational metrics can indirectly support security by enabling rapid response to staffing issues that might impact service levels or security incident response capabilities. Data privacy for scheduling information should be maintained. *Terraform Support*: Supported for Amazon Connect instance management via Terraform AWS Provider: aws_connect_instance in v3.50.0+ (this specific feature is an internal operational enhancement, not a configurable parameter via IaC).
Hybrid Cloud Solutions
- Amazon Elastic VMware Service (EVS) – i7i.metal-48xl Instance Support: Amazon EVS now supports the i7i.metal-48xl Amazon EC2 bare-metal instance type, offering a higher core-count option with a newer generation processor. Powered by 5th generation Intel Xeon Scalable processors, it delivers up to 23% better compute performance and more than 10% better price performance over i4i instances. This provides more powerful underlying hardware for VMware environments, potentially consolidating workloads and simplifying security patching and configuration management. Ensuring secure network connectivity between EVS and on-premises environments is critical for hybrid cloud deployments. *Terraform Support*: Terraform manages supporting infrastructure (e.g., networking, IAM, KMS) for EVS. The EVS host type selection is typically configured within the EVS console or API. Terraform AWS Provider Documentation in v4.65.0+.
Storage & Data Protection
- AWS Backup – Cross-Account/Region Copy for FSx for NetApp ONTAP: AWS Backup now enables copying Amazon FSx for NetApp ONTAP backups to a different AWS Region, a different AWS account, or both. This functionality can be configured through policy-based backup plans or initiated via on-demand copy jobs. This is a critical Disaster Recovery (DR) and resilience feature, significantly enhancing data protection and compliance. Cross-account/cross-Region backups are a cornerstone of robust DR strategies, protecting against regional outages and insider threats/account compromise. Implementing this via policy ensures consistent application of DR controls, and KMS encryption for backups is essential. *Terraform Support*: Supported for backup plan configuration via Terraform AWS Provider: aws_backup_plan (using the
copy_actionblock) in v2.0.0+.
Analytics & Data Governance
- AWS Glue 5.1 – Engine Upgrades & Lake Formation FGAC for Writes: AWS Glue 5.1 upgrades core engines to Apache Spark 3.5.6, Python 3.11, and Scala 2.12.18, bringing performance and security enhancements. It also introduces support for Apache Iceberg 3.0 and extends AWS Lake Formation fine-grained access control (FGAC) to write operations (DML and DDL) for Spark DataFrames and Spark SQL, as well as full-table access control for Apache Hudi and Delta Lake tables. This is a significant data governance and security update. Extending Lake Formation FGAC to DML/DDL operations is a game-changer for data lake security, ensuring that only authorized identities can modify data, not just read it. This is crucial for compliance (e.g., SOX, GDPR) and preventing data integrity issues. Row lineage tracking in Iceberg 3.0 also aids in auditability. *Terraform Support*: Supported for Glue job configuration via Terraform AWS Provider: aws_glue_job (setting
glue_version = "5.1") in v4.0.0+. Lake Formation permissions are managed viaaws_lakeformation_permissions.
Announcements in this summary
- Amazon Bedrock AgentCore Memory Generative AI
- Amazon EC2 Compute
- AWS Transform Migration and Modernization
- Amazon EC2 Compute
- Amazon Bedrock AgentCore AI/ML
- Amazon Aurora MySQL Database
- Amazon CloudWatch Monitoring
- Amazon Bedrock Machine Learning
- Amazon SageMaker Machine Learning
- Amazon SageMaker Machine Learning
- Amazon Connect Contact Center
- Amazon Elastic VMware Service Hybrid Cloud
- AWS Backup Storage
- AWS Glue Analytics
- Amazon Aurora DSQL Database