Weekly Cloud Info #W33 - 2025

Author

Walid LARABI
August 15, 2025

Hi!

Explore AWS's new Advanced Go Driver, Cloud Control API, and EC2 P5.4xlarge instance. Google Cloud now supports OpenAI's gpt-oss models on GKE. Azure introduces Tenant-Level Service Health Alerts and enhanced app testing capabilities.

Have a great read. ☕

📰 Top picks of the week

AWS Advanced Go Driver Now Available for RDS and Aurora with Enhanced Failover Features

The AWS Advanced Go Driver is now generally available for Amazon RDS and Amazon Aurora PostgreSQL and MySQL-compatible databases. It enhances switchover and failover times, supports Federated Authentication, and integrates with AWS Secrets Manager and IAM. The driver, which wraps the pgx PostgreSQL and native MySQL drivers, reduces writer failover times to single-digit seconds and is available as an open-source project under the Apache 2.0 license.

AWS Introduces Cloud Control API MCP Server for Natural Language Infrastructure Management

AWS has launched the Cloud Control API (CCAPI) MCP Server, enabling developers to manage AWS infrastructure using natural language commands for CRUDL operations. This tool simplifies infrastructure management and accelerates onboarding by translating conversational inputs into cloud actions. Developers can configure the server using a JSON file and set AWS credentials, with options for read-only operations.

Amazon EKS Adds Cilium Support for Enhanced Hybrid Node Networking

Amazon Elastic Kubernetes Service (EKS) now supports Cilium as the Container Networking Interface (CNI) for Amazon EKS Hybrid Nodes. This integration allows users to utilize advanced features such as application ingress, in-cluster load balancing, Kubernetes network policies, and kube-proxy replacement mode, all powered by Cilium's eBPF technology. Support is available in all AWS Regions where EKS Hybrid Nodes are offered.

AWS Launches Cost-Effective EC2 P5.4xlarge Instance with NVIDIA H100 GPU

AWS has introduced the new Amazon EC2 P5.4xlarge instance, featuring one NVIDIA H100 GPU, designed for cost-effective machine learning and high-performance computing. This instance allows businesses to scale their resources flexibly, making it suitable for developing small to medium Large Language Models and deploying HPC applications. The P5 instances are available in multiple AWS regions and can be purchased through various pricing options.

AWS DynamoDB Increases Throughput Mode Update Limit to Four Times Daily

Amazon DynamoDB now allows users to update their table throughput mode from provisioned capacity to on-demand mode up to four times within a 24-hour period. This change facilitates easier data loading and workload management, enhances CloudFormation deployments, and supports serverless applications with automatic scaling and pay-per-request pricing. There are no charges for switching capacity modes, and the update applies to all DynamoDB tables across all AWS Regions.

Google Cloud Adds Support for OpenAI's gpt-oss Models on GKE

Google Cloud now supports the deployment of OpenAI's new gpt-oss models (gpt-oss-120b and gpt-oss-20b) on Google Kubernetes Engine (GKE), providing detailed benchmarks for optimal performance. The GKE Inference Quickstart tool simplifies the deployment process with pre-built configurations, allowing users to efficiently serve these large models using NVIDIA GPUs. Customers can also utilize the gcloud CLI for deployment, enhancing accessibility to advanced AI capabilities.

GKE Enhances Container Image Streaming for Faster Startup Times

Google Kubernetes Engine (GKE) has introduced performance improvements to its container image streaming feature, enhancing application startup times by up to 30%. Key updates include intelligent read-ahead capabilities that proactively fetch image data, along with backend optimizations for efficient data serving. These enhancements particularly benefit AI/ML model serving applications with large container images, leading to faster container startup and better resource utilization.

Azure Monitor Launches Tenant-Level Service Health Alerts for Enhanced Incident Response

Azure Monitor has introduced Tenant-Level Service Health Alerts in preview, allowing customers to receive proactive notifications about service health issues that affect their entire tenant. Users can create alert rules at the directory level from the Service Health page or the alert rule creation wizard in the Azure portal. This feature enhances visibility and response times for incidents impacting tenant-scoped services, and it is recommended to configure both subscription-level and tenant-level alerts for comprehensive coverage.

Azure App Testing Integrates Load Testing and Playwright for Enhanced Application Testing

Azure App Testing now allows developers and QA teams to conduct large-scale functional and performance tests using frameworks like Playwright, JMeter, and Locust. This feature combines Azure Load Testing (now generally available) and Playwright Workspaces (currently in public preview) into a single hub within the Azure Portal, streamlining resource provisioning, access control, and billing processes.

Azure App Service Introduces Inbound IPv6 Support for Multi-Tenant Apps

Inbound IPv6 support is now generally available for public multi-tenant App Service across all Azure regions, applicable to multi-tenant apps on Basic, Standard, and Premium SKUs, as well as Functions Consumption, Functions Elastic Premium, and Logic Apps Standard.

❤️ You might also like

  • Perplexity offers $34.5 billion to acquire Google Chrome amid antitrust proceedings LINK

  • OpenAI’s open-weight models (gpt-oss-120B & 20B) now available on AWS via Bedrock and SageMaker LINK

  • Oracle and Google partner to integrate Gemini AI models into Oracle Cloud and enterprise apps LINK

  • NTT DATA teams up with Google Cloud to deploy agentic AI and cloud modernization solutions globally LINK

  • Google pledges $9 billion investment in AI and cloud infrastructure in Oklahoma LINK

  • U.S. GSA launches "USAi", a secure cloud-based platform offering government access to AI tools (OpenAI, Anthropic, Google, Meta) LINK

🏁 Enjoy this newsletter?

Forward it to a friend, and let them know they can subscribe here.