Announcing Valkey 9.1 for Amazon ElastiCache

Mastering Secure CI/CD for ECS with GitHub Actions

Bài viết hướng dẫn xây dựng quy trình CI/CD an toàn cho ECS trên Fargate bằng GitHub Actions, tập trung vào tối ưu vận hành (gộp workflows, dùng commit hash cho image tag, quản lý task definitions/services bằng ecspresso) và tăng cường bảo mật (AssumeRole không khóa tĩnh bằng OIDC, quét lỗ hổng với Trivy/Dockle, multi-stage builds). Mục tiêu là giúp người đọc triển khai pipeline vừa hiệu quả vừa an toàn cho dự án.

Lập trình viên cần đọc bài này để tìm hiểu cách xây dựng và tối ưu hóa chuỗi giao tiếp liên tục (CI/CD) trên AWS ECS với GitHub Actions, từ đó nâng cao bảo mật, giảm rủi ro và streamline quá trình phát triển ứng dụng.

Accelerating government modernization with AWS

Atlassian and AWS have expanded their Strategic Collaboration Agreement to support government technology modernization. The partnership offers public sector agencies migration paths from Atlassian Data Center to Atlassian Cloud or FedRAMP-Moderate-authorized Atlassian Government Cloud, both built on AWS infrastructure. Key benefits include improved cross-agency collaboration, AI capabilities, enterprise-grade security features like customer-managed keys and Guard Premium, and compliance support for federal requirements.

Amazon S3 server access logs now deliver to Amazon CloudWatch Logs and Amazon S3 Tables

Amazon S3 now supports delivering server access logs to two new destinations: Amazon CloudWatch Logs and Amazon S3 Tables. CloudWatch Logs delivery enables instant querying, alarms, cross-account and cross-Region aggregation, and KMS encryption. S3 Tables mirroring stores logs in Apache Iceberg format at no additional storage cost, making them immediately queryable via SQL in Athena, Redshift, and other Iceberg-compatible engines. These options complement the existing free delivery to S3 general purpose buckets. The feature is available in all AWS Regions except China and GovCloud.

Building Financial Hierarchies with Amazon Neptune for Treasury Operations

Amazon's FinTech team shares how they use Amazon Neptune to model complex corporate treasury structures as a property graph. The post covers an iterative graph modeling approach that evolved through five steps — from basic directed edges to labeled edges, bidirectional relationships, node-level eligibility properties, and bank account vertices. Key patterns include Gremlin queries for POBO payment routing, cash sweep path resolution, and bank account eligibility lookups. A Primary/State node versioning pattern preserves historical hierarchy states for audit and regulatory compliance. The architecture pairs Neptune (for graph traversal) with DynamoDB (for reference data), using API Gateway caching to reduce load. Lessons learned cover graph model design, property placement strategy, Gremlin performance tuning, and extensibility for future hierarchy types like legal entity structures and bank-to-bank relationships.

Amazon EMR Serverless now supports live configuration updates without application restarts

Amazon EMR Serverless giờ đây hỗ trợ cập nhật cấu hình ứng dụng (như dung lượng tối đa hoặc image tùy chỉnh) theo thời gian thực mà không cần khởi động lại ứng dụng. Các tác vụ mới sau cập nhật sẽ sử dụng cấu hình mới, trong khi tác vụ đang chạy vẫn hoạt động bình thường.

Lập trình viên xử lý dự án big data trên EMR Serverless sẽ tiết kiệm thời gian và chi phí khi có thể cập nhật cấu hình trực tiếp mà không cần dừng lại, giữ cho công việc chạy liên tục mà không cần restart ứng dụng.

AWS GPU prices jump 20% as memory crunch bites

AWS has raised EC2 Capacity Blocks for ML prices by roughly 20% starting July, marking the second hike in six months after a 15% increase in January. The increases target reserved Nvidia GPU blocks used by AI teams for large model training and fine-tuning, while other purchasing options and Amazon's own Trainium chip remain unaffected. The root cause is a high-bandwidth memory shortage that is constraining GPU production and data center capacity globally. The scarcity gives hyperscalers like AWS, Microsoft, and Google pricing power since customers have few alternatives. The same memory crunch is driving up prices across Apple hardware and Xbox, while benefiting memory makers like Micron and SK Hynix. AI teams relying on reserved compute now face rising and unpredictable reservation costs.

Unified Secrets Security with GitGuardian and AWS Secrets Manager

Organizations using AWS Secrets Manager still face a visibility gap — they can't see which vaulted secrets have leaked into code, CI/CD logs, or AI tool configurations. GitGuardian's ggscout collector bridges this gap by cataloging secrets from Secrets Manager and correlating them with exposed credentials found across repositories using a cryptographic HMSL protocol, so actual secrets never leave your AWS environment. The integration enables five key capabilities: detecting vaulted secret exposures in code, prioritizing incident response by severity, identifying secret sprawl across multi-account AWS architectures, tracking remediation progress, and enforcing continuous rotation governance policies. A phased implementation roadmap covers deployment (EC2, Docker, or EKS), initial assessment, exposure analysis, automation of scheduled scans and alerts, and ongoing KPI monitoring.