Greetings Cloud Security Architects, DevSecOps Leads, and Terraform Specialists,
Today's AWS update brings a suite of enhancements across AI/ML, networking, security, and operational governance, reinforcing AWS's commitment to innovation, resilience, and developer agility. We'll dive into each feature, dissecting its implications for security, architecture, and compliance, alongside crucial insights into its current Terraform availability.
Security, Identity, and Compliance
- Amazon Cognito Multi-Region Replication (MRR) in AWS GovCloud (US)
- Feature & Security Summary: Amazon Cognito MRR now extends its robust resilience capabilities to AWS GovCloud (US) Regions. This feature enables near real-time synchronization of user and machine identity data, including credentials and user pool configurations, to a designated standby user pool in a secondary region. This is a critical enhancement for business continuity and disaster recovery, ensuring authentication availability even during primary region disruptions. From a security and compliance perspective, this is paramount for highly regulated environments like GovCloud, allowing organizations to meet stringent RTO/RPO objectives for identity management and maintain data residency requirements by keeping identity data within the GovCloud boundary. Architectural considerations include designing for multi-region failover and ensuring proper KMS key management for the required multi-Region customer-managed keys.
- *Terraform Support*: Supported in Terraform AWS Provider: aws_cognito_user_pool and Terraform AWS Provider: aws_kms_key (for
multi_regionKMS keys) in v5.15.0+.
- AWS Continuum for Code Vulnerabilities
- Feature & Security Summary: AWS Continuum is a groundbreaking DevSecOps offering designed to discover, validate, prioritize, and remediate code vulnerabilities at machine speed. It builds a comprehensive context graph of your AWS environment, identifying exploitable findings by testing them in an isolated sandbox. This service then prioritizes vulnerabilities based on business impact and provides remediation patches as pull requests directly into developer workflows, complete with exploitability proof and blast radius assessments. Continuum significantly elevates the security posture by automating the vulnerability lifecycle, reducing mean time to remediation (MTTR), and focusing on real business risks. It operates in "Learn" mode (proposing fixes) or "Enforce" mode (autonomous action within guardrails), integrating seamlessly with Amazon GuardDuty and AWS Security Hub. This is a critical tool for achieving continuous security and compliance in modern cloud-native development.
- *Terraform Support*: Pending provider release (use AWS CLI/SDK or custom provisioner in the interim).
Machine Learning & AI
- Amazon SageMaker AI: Training Session APIs for Serverless Model Customization
- Feature & Security Summary: Amazon SageMaker AI now supports training session APIs for serverless model customization, providing developers with granular control over post-training algorithms without the burden of GPU infrastructure management. This allows teams to write custom training loops using operations like
forward_backward(),optim_step(), andsample()against a live stateful session. SageMaker handles the underlying GPU allocation, model/LoRA loading, distributed execution, fault recovery, checkpoint storage, and security. This enhances developer agility and reduces operational overhead, while security benefits from SageMaker's managed environment, including encrypted checkpoint storage and fault recovery mechanisms. Billing is token-based, optimizing cost for local computation. - *Terraform Support*: Pending provider release (use AWS CLI/SDK or custom provisioner in the interim).
- Amazon SageMaker AI: Serverless Customization for New Foundational Models
- Feature & Security Summary: Expanding its serverless model customization capabilities, Amazon SageMaker AI now supports Muse Glimmer, Qwen3.8 27B, and Ministral 3 3B, 8B, and 14B models. This enables customers to adapt these foundational models to their specific domains and workflows using techniques like supervised fine-tuning (SFT), direct preference optimization (DPO), reinforcement fine-tuning (RFT), low-rank adaptation (LoRA), or full fine-tuning. SageMaker AI continues to abstract away all infrastructure provisioning and training orchestration, allowing ML teams to focus purely on model adaptation. This broadens the scope for specialized AI applications, with security benefits derived from SageMaker's managed and secure training environments.
- *Terraform Support*: Pending provider release (use AWS CLI/SDK or custom provisioner in the interim).
- Amazon Bedrock AgentCore Gateway: InvokeModel API Support and Fallback Routing
- Feature & Security Summary: The Amazon Bedrock AgentCore Gateway now supports the Amazon Bedrock
InvokeModelandInvokeModelWithResponseStreamAPIs on its inference endpoint, alongside a crucial model fallback routing capability. This allows existing Bedrock SDK integrations to point to a single, governed gateway endpoint for accessing any model provider, eliminating the need for client-side code rewrites for each provider. The model fallback feature significantly enhances resilience by automatically routing requests to alternative models or providers when the primary destination experiences throttling, capacity issues, or unreachability. Administrators define an ordered chain of fallback destinations and trigger conditions, with a built-in circuit breaker to skip failing destinations. This is a critical security and resilience enhancement, centralizing AI model access for improved governance, observability, and robust application availability in multi-model and multi-provider AI architectures. - *Terraform Support*: Pending provider release (use AWS CLI/SDK or custom provisioner in the interim).
Networking
- AWS Interconnect - Last Mile Preview with Megaport
- Feature & Security Summary: AWS has launched AWS Interconnect, a last-mile preview in collaboration with Megaport, offering a fully managed connectivity solution. This service simplifies connecting branch offices, data centers, and remote locations to AWS with minimal configuration. Customers can select their preferred AWS Region, bandwidth, Direct Connect Gateway ID, and partner subscriber ID via the AWS Management Console, receiving an activation key for Megaport provisioning. The service automates BGP peering, VLAN configuration, and ASN assignment, supporting dynamic bandwidth scaling from 1 Gbps to 100 Gbps. This significantly reduces the friction and complexity of hybrid cloud network setup, enhancing secure, private connectivity, and offering high availability with a 99.99% SLA.
- *Terraform Support*: Pending provider release (use AWS CLI/SDK or custom provisioner in the interim).
- AWS Direct Connect: 100 Gbps Connections in Bangalore, India
- Feature & Security Summary: AWS Direct Connect has expanded its 100 Gbps dedicated connection offerings at the Netmagic DC2 data center near Bangalore, India. This enables customers in the region to establish high-bandwidth, private, and direct network access to all public AWS Regions (excluding China), AWS GovCloud Regions, and AWS Local Zones. This expansion is vital for organizations requiring massive data transfers, low-latency connectivity, and enhanced network security between their on-premises infrastructure and AWS cloud resources in the Indian subcontinent. It supports demanding workloads and large-scale hybrid cloud deployments.
- *Terraform Support*: Supported in Terraform AWS Provider: aws_dx_connection (argument
bandwidthfor100Gand specificlocationfor Bangalore) in v2.0.0+.
Serverless Compute
- AWS Lambda MicroVMs: Expanded Regional Availability
- Feature & Security Summary: AWS Lambda MicroVMs, a serverless compute primitive offering VM-level isolation, near-instant launch/resume speeds, and state preservation, is now available in four additional AWS Regions: Asia Pacific (Hong Kong), Asia Pacific (Malaysia), Asia Pacific (Thailand), and Europe (Spain). This expansion provides customers in these regions with a highly secure and performant option for stateful serverless applications. MicroVMs enhance security by providing strong isolation boundaries between workloads, while their rapid launch times and state preservation capabilities are ideal for latency-sensitive and persistent applications.
- *Terraform Support*: The underlying AWS regions are supported by the Terraform AWS Provider in v5.50.0+. While the Lambda MicroVMs service is available in these regions, explicit Terraform resources or arguments for configuring MicroVMs specifically are pending provider release.
Search and Analytics
- Amazon OpenSearch Serverless: Reduced Default Refresh Interval
- Feature & Security Summary: Amazon OpenSearch Serverless has reduced its default refresh interval to 5 seconds for both search and vector collections, down from 10 seconds. This change, applied automatically without any configuration required, means newly ingested documents become searchable twice as fast. This significantly improves read-after-write visibility, benefiting applications that rely on near real-time data for analytics, monitoring, or operational dashboards. While not a direct security feature, faster data availability can enhance the responsiveness of security monitoring and threat detection systems that leverage OpenSearch Serverless for log analysis.
- *Terraform Support*: This is a service-level default change and does not expose a direct configuration parameter for Terraform.
- Amazon Quick: Now Available in Asia Pacific (Singapore) Region
- Feature & Security Summary: Amazon Quick, the AI-powered agentic teammate for business users, is now available in the AWS Asia Pacific (Singapore) region (ap-southeast-1). This expansion allows customers in Singapore to leverage Quick's AI-powered chat, Research, Spaces, and QuickSight dashboards while adhering to local data sovereignty requirements, as data is stored and processed locally within the region. This is a key development for organizations with strict data residency policies, enabling them to empower their workforce with AI-driven insights and actions without compromising compliance.
- *Terraform Support*: The AWS Asia Pacific (Singapore) region is supported by the Terraform AWS Provider in v4.60.0+. Specific Terraform resources for Amazon Quick are pending provider release.
Management and Governance
- AWS CloudFormation: Force Rollback Option for Continue Update Rollback
- Feature & Security Summary: AWS CloudFormation now offers a
force rollbackoption for itscontinue update rollbackoperation. This enhancement allows CloudFormation to complete an update rollback and return a stack to a deployable state in a single operation, even when certain resources are stuck or require external attention. After a forced rollback, users can create a drift-aware change set to review any remaining configuration differences. This feature significantly improves the operational resilience and recovery capabilities of CloudFormation deployments, reducing the time and complexity associated with resolving failed stack updates. While not a direct security control, improved operational stability and faster recovery from deployment failures are critical for maintaining a secure and compliant infrastructure-as-code posture. - *Terraform Support*: This is an operational command for CloudFormation stacks, not a configurable resource attribute. While Terraform manages CloudFormation stacks, this specific operational capability is typically invoked via AWS CLI/SDK.
Announcements in this summary
- Amazon OpenSearch Serverless Search and Analytics
- Amazon SageMaker AI Machine Learning
- AWS CloudFormation Management and Governance
- Amazon Connect Customer Profiles contact-center
- AWS Continuum Security
- AWS Interconnect Networking
- Amazon SageMaker AI Machine Learning
- Amazon Bedrock AgentCore Gateway AI/ML
- AWS Direct Connect Networking
- AWS Lambda MicroVMs Serverless Compute
- Amazon Quick Analytics
- Amazon Cognito Security, Identity, and Compliance