All news in category "Vendor and Hyperscaler Watch"

AWS Launches General Availability of EC2 R8gn Instances

🚀 AWS has made the new Amazon EC2 R8gn instances generally available, built on AWS Graviton4 processors to deliver up to 30% better compute performance than Graviton3. R8gn instances include 6th generation Nitro Cards and offer up to 600 Gbps network bandwidth—the highest among network-optimized EC2 instances—and sizes up to 48xlarge with metal options. They provide up to 1,536 GiB memory, up to 60 Gbps EBS bandwidth, and support Elastic Fabric Adapter (EFA) on select large and metal sizes for lower latency cluster performance.

On-demand deployment for custom Meta Llama models on Bedrock

🚀 Amazon Bedrock now offers an on-demand deployment option for customized Meta Llama 3.3 models that have been fine-tuned or distilled in Bedrock; models customized on or after September 15, 2025 are eligible. The feature lets customers process requests in real time and pay only for consumed compute, removing the need for pre-provisioned always-on resources. Bedrock continues to provide a managed platform with built-in security, privacy, and responsible AI capabilities.

Amazon Managed Service for Prometheus Adds 11 Regions

📢 Amazon Managed Service for Prometheus is now generally available in 11 additional AWS regions, including Asia Pacific (Jakarta, Hyderabad, Osaka, Melbourne, Taipei), Canada West (Calgary), Europe (Spain), Israel (Tel Aviv), Mexico (Central), Middle East (Bahrain), and US West (N. California). The fully managed, Prometheus-compatible monitoring service makes it easier to collect, store, query, and alarm on operational metrics at scale. Customers can send up to 1 billion active metrics to a single workspace and create multiple workspaces per account to partition workloads. See the AWS user guide or product documentation for the full list of supported regions and configuration details.

Google releases XProf and Cloud Diagnostics XProf tools

🔧 Google has open-sourced XProf, an upgraded ML profiler, and published the Cloud Diagnostics XProf library to simplify profiling and optimizing models on xPUs. The release brings unified XLA-based profiling across JAX, PyTorch/XLA and TensorFlow/Keras, and supports programmatic and on-demand trace capture. The Cloud Diagnostics library packages dependencies, stores profiles in Google Cloud Storage for retention, provisions TensorBoard on VMs or GKE for faster loading, and produces shareable links for collaborative analysis with tunable machine types for performance.

Microsoft warns Windows 10 support ends in 30 days

⚠️ Microsoft reminded customers that Windows 10 will reach end of servicing on October 14, 2025, with the October monthly update being the last security release for affected versions. After that date, Microsoft will no longer provide bug fixes or technical assistance for security, stability, or usability issues. Customers are advised to upgrade eligible devices to Windows 11, migrate to Windows 365 in the cloud, enroll in Extended Security Updates (ESU), or consider LTSC/LTSC alternatives for specialized devices.

ECS Service Connect Enables Cross-Account Namespace Sharing

🚀 Amazon ECS Service Connect now supports cross-account communication by letting teams share AWS Cloud Map namespaces via AWS RAM. Platform engineers can create a resource share and grant access to individual accounts, OUs, or the whole organization so services in multiple accounts register to a single namespace. This reduces duplication, simplifies service discovery, and works with Fargate and EC2 across commercial regions.

AWS Direct Connect Adds Support for 4-Byte AS Numbers

🌐 AWS Direct Connect now supports 4-byte Autonomous System (AS) numbers for virtual interfaces, extending the usable ASN range to the full RFC 6793 limit of 4,294,967,294. The change addresses constraints imposed by the legacy 2-byte ASN space (max 65,536), which can complicate complex or multi-tenant network topologies. Support is available in all AWS regions and across all Direct Connect virtual interface types. To provision, use the Direct Connect Console or the updated APIs to specify 4-byte ASNs when creating virtual interfaces.

AlloyDB on Axion C4A Instances Reaches General Availability

🚀 AlloyDB on Axion-powered C4A virtual machines is now generally available. AlloyDB on C4A offers substantial efficiency gains—up to 45% better price-performance versus N‑series VMs and up to 2× higher throughput versus Amazon Aurora on Graviton4—targeting transactional and data‑intensive workloads. C4A supports shapes from 1 to 72 vCPUs (including a new 48‑vCPU option and a cost‑effective 1‑vCPU/8GB shape for dev/sandbox). Deploy or upgrade C4A instances via the AlloyDB console in supported regions.

Windows 11 23H2 Home and Pro end support Nov 11, 2025

⚠️ Microsoft warned that devices running Windows 11 23H2 Home and Pro editions will reach end of servicing on November 11, 2025, with the November 2025 monthly security update as the last release for those editions. Enterprise and Education SKUs will continue to receive mainstream support until November 10, 2026. Users are advised to upgrade to Windows 11 24H2, but Microsoft has applied safeguard holds for systems with incompatible Intel Smart Sound Technology audio drivers, SenseShield code‑obfuscation drivers, wallpaper customization tools, certain integrated cameras, and Dirac audio software.

Rent the Runway Boosts Developer Velocity with Cloud SQL

🚀 Rent the Runway migrated from a complex, self-managed MySQL environment to Cloud SQL, gaining managed backups, built-in observability, and simplified disaster recovery. The move removed a third-party 24/7 MySQL support vendor and cut more than $180,000 in annual operational costs. Engineers now work in IAM-controlled instances with query insights, enabling safer schema changes, CI/CD for database updates, and faster, more reliable deployments.

OpenTelemetry (OTLP) Support Added to Cloud Trace UI

🚀 Cloud Trace now accepts traces via OTLP through telemetry.googleapis.com, enabling vendor-agnostic ingestion and preserving the OpenTelemetry data model. Using OTLP reduces client-side exporter complexity, improves telemetry integrity, and enables easier routing to other backends. The Trace Explorer leverages OpenTelemetry semantic conventions for improved filtering and diagnostics. Google recommends telemetry.googleapis.com as the best practice for new and high-volume trace pipelines and provides a migration guide to help users transition.

AWS EC2 M4 and M4 Pro Mac instances now generally available

🚀 Amazon EC2 M4 and M4 Pro Mac instances are now generally available, offering up to 20% (M4) and 15% (M4 Pro) improved application build performance versus the M2 families. Powered by the AWS Nitro System, they provide up to 10 Gbps network and 8 Gbps EBS bandwidth and include a 2 TB instance store per Dedicated Host. M4 models use Apple M4 Mac Mini hardware and support macOS Sequoia 15.6+. Available in US East (N. Virginia) and US West (Oregon).

Three Critical Needs for Responding to a Cyberattack

🛡️ When a cyberattack strikes, three capabilities matter most: clarity, control and a dependable lifeline. Real-time visibility and a unified view enable quick detection of anomalies and identification of the blast radius. Rapid containment—isolating endpoints, revoking credentials and enforcing policies—prevents lateral spread. Finally, immutable backups, granular restores and orchestrated disaster recovery ensure fast, trusted restoration of operations.

Amazon SageMaker Adds EC2 P6-B200 Notebook Instances

🚀 Amazon Web Services announced general availability of EC2 P6-B200 instances for SageMaker notebooks. These instances include eight NVIDIA Blackwell GPUs with 1,440 GB of high-bandwidth GPU memory and 5th Gen Intel Xeon processors, offering up to 2x the training performance versus P5en. They enable interactive development and fine-tuning of large foundation models in JupyterLab and CodeEditor, and are available in US East (Ohio) and US West (Oregon).

AWS Adds Five EFA Metrics to Improve Network Observability

🔍 AWS has introduced five new Elastic Fabric Adapter (EFA) metrics to improve network observability for AI/ML and HPC workloads. The counters track retransmitted packets and bytes, retransmit timeouts, impaired remote connections, and unresponsive remote receivers at the per-EFA device level. Available on Nitro v4+ instances with EFA installer 1.43.0+, metrics are exposed via sysfs and can be exported to Prometheus and tools like Grafana for monitoring and alerting.

AWS GuardDuty S3 Malware Scanning Now Handles Larger Files

🛡️ AWS has expanded GuardDuty Malware Protection for S3 scanning limits, raising the maximum file size from 5 GB to 100 GB and increasing archive processing to 10,000 files per archive (previously 1,000). These enhancements are automatically enabled in all supported AWS Regions. Customers gain broader coverage for large objects and dense archives stored in S3, improving pre-ingestion threat detection. This update strengthens protection for workloads and downstream processes.

AWS Adds Deadline Cloud Availability in Seoul and London

🎬 AWS Deadline Cloud is now available in Asia Pacific (Seoul) and Europe (London). This fully managed service simplifies render management for teams producing computer-generated graphics and visual effects, enabling them to scale render farms near creative teams. Customers gain improved integration with existing AWS services and creative pipelines, and can now deploy Deadline Cloud across 10 AWS regions worldwide. Bringing the service closer reduces latency and streamlines collaboration across distributed production workflows.

Amazon RDS Proxy Adds End-to-End IAM Authentication

🔐 Amazon RDS Proxy now supports end-to-end IAM authentication for Amazon Aurora and RDS database instances, allowing applications to authenticate through the proxy using AWS IAM without storing credentials in Secrets Manager. This reduces credential rotation overhead and simplifies credential management. The capability is available for MySQL and PostgreSQL in all Regions where RDS Proxy is supported.

SageMaker Unified Studio Connects Remotely to VS Code

🔗 AWS now enables remote connections from local VS Code to Amazon SageMaker Unified Studio, allowing developers to use their personalized VS Code setups while running workloads on SageMaker-managed compute and accessing cloud-resident data. Authentication is provided via the AWS Toolkit extension for secure, streamlined access. The integration preserves existing development workflows for data processing, SQL analytics, and ML.

Microsoft adds malicious link warnings to Teams chats

🔔 Microsoft Teams will display warnings on private messages that contain URLs flagged as spam, phishing, or malware for customers using Microsoft Defender for Office 365 and enterprise Teams. The feature enters public preview for desktop, Android, web, and iOS in September 2025 and is slated for general availability in November 2025. Admins can enable the preview via the Teams Admin Center messaging settings; warnings will be enabled by default at GA and can be managed through the Teams Admin Center or PowerShell.