All news with #aws tag

📊 Amazon Connect now exposes eight new metrics to evaluate AI agent outcomes, including goal success rate, faithfulness score, and tool selection accuracy. These metrics give contact center teams visibility into whether AI-driven interactions resolve customer requests and where contextual hallucinations occur. Metrics are accessible from the AI Agent Performance dashboard, the GetMetricDataV2 API, or a zero-ETL data lake for custom reporting. This capability is available in all Regions that support Amazon Connect AI Agents.

🚀 AWS has added High Memory U7i instances in new regions, bringing the u7i-8tb.112xlarge to Europe (Stockholm, Zurich), the u7in-16tb.224xlarge to US East (Ohio), and the u7in-24tb.224xlarge to Europe (Stockholm). These 7th-generation instances are powered by custom 4th-generation Intel Xeon Scalable processors (Sapphire Rapids) and DDR5 memory, offering 8, 16, and 24 TiB options with 448 or 896 vCPUs. They deliver high EBS and network throughput (up to 100 Gbps EBS; up to 100–200 Gbps network) and include ENA Express, making them well suited for mission-critical in-memory databases such as SAP HANA, Oracle, and SQL Server.

🏦 AWS Marketplace sellers can now delete bank accounts directly from the Payment Settings page in the AWS Marketplace Management Portal (AMMP). This self-service capability supports removal of ACH- and SWIFT-type accounts without contacting customer service, and shows Last Updated timestamps for each entry. The change helps enterprises and ISVs clean up unused or failed accounts, reduce payment routing risk, and simplify multi-currency banking management.

⚙️ AWS Deadline Cloud now supports configurable pre- and post-submission scripts, enabling studios to run custom pipeline logic as part of every job submission. Pre-submission hooks can validate job configurations, discover and attach additional inputs such as textures or caches, modify submission parameters, or enforce studio policies before uploads; post-submission hooks can send notifications, update tracking systems, or log submission details. Scripts are defined in simple YAML or JSON files placed per-job or in a shared directory via an environment variable, receive job metadata automatically, and support configurable timeouts.

🔗 Amazon Athena now provides managed connectors for 12 external data sources, including DynamoDB, PostgreSQL, MySQL, and Snowflake, enabling queries against data outside Amazon S3 without deploying connector infrastructure. Athena creates and manages AWS Glue Data Catalog federated connectors on your behalf and registers each source as a federated catalog. You can query those sources alongside S3 data and optionally apply fine‑grained access controls through AWS Lake Formation. Federated queries are available in all standard AWS Regions except AWS GovCloud (US) and China Regions.

🔗 AWS announced native integration between AWS Client VPN and AWS Transit Gateway, enabling centralized remote access across multiple VPCs and on-premises networks without an intermediate VPC. Client source IPs are preserved end-to-end, allowing authorization rules and forensic tracing to map traffic back to specific users. Transit Gateway flow logs capture connection-level details tied to those preserved client IPs, improving troubleshooting and auditability. The integration is available in all Regions where Client VPN is offered and incurs no additional charges beyond standard service pricing.

🚀 AWS Parallel Computing Service (AWS PCS) now supports Slurm 25.11, including an expedited re-queue feature that can automatically reschedule jobs affected by node issues at highest priority. You can enable a Prometheus-compatible OpenMetrics endpoint for real-time visibility into jobs, nodes, and scheduling using existing monitoring tools. AWS PCS can forward slurmdbd and slurmrestd logs to Amazon CloudWatch Logs, Amazon S3, or Amazon Data Firehose. Scheduler audit logs are now delivered as a dedicated log type to provide independent control over ingestion, retention, and storage costs.

🔧 Amazon SageMaker HyperPod now automatically selects and continuously maintains the optimal network topology for Slurm GPU clusters based on the instance types in the cluster. By choosing tree or block topology models that match instance interconnect characteristics, HyperPod reduces GPU-to-GPU latency, improves NCCL collective efficiency, and raises distributed training throughput. The topology adapts automatically during scale-up, scale-down, and node replacement events, so administrators no longer need to manually edit topology files or reconfigure Slurm. Topology-aware scheduling is enabled by default across supported AWS Regions.

🚀 Amazon SageMaker Unified Studio now supports serverless notebooks and a built-in data agent for AWS IAM Identity Center (IdC) domains, extending functionality that was previously available only in IAM domains. The serverless notebook provides a single interactive workspace for SQL, Python, large-scale data processing, ML workloads, and visualizations. A built-in AI data agent generates code and SQL from natural-language prompts and helps guide users through tasks. The environment is backed by Amazon Athena for Apache Spark, scaling from interactive queries to petabyte-scale processing, and is available in all Regions where Unified Studio is supported.

🔧 Amazon has released the aws-smus-cicd-cli, an open-source command-line tool that automates deployment of multi-service data and AI applications built in SageMaker Unified Studio. Teams define applications once in an manifest.yaml, and the CLI substitutes stage-specific settings, provisions resources in dependency order, and runs post-deployment tests. Four lifecycle commands — describe, bundle, deploy, test — integrate with existing CI/CD pipelines to reduce configuration drift and speed releases. The CLI is available at no additional cost in supported Regions; you pay only for the AWS resources provisioned during deployment.

⚙️ AWS Compute Optimizer now supports the newest EC2 and RDS instance types, expanding recommendations to include Compute, General Purpose, Memory-optimized, Memory-intensive, and Storage-optimized families. The update covers EC2 families such as C8*, M8*, R8*, x8i, and i7i, and RDS classes including M7i, M8g, R8g, X1, and Z1d across MySQL, PostgreSQL, Amazon Aurora MySQL, and Aurora PostgreSQL. This capability is available in all standard AWS Regions where Compute Optimizer operates, excluding AWS GovCloud (US) and the two China regions.

🔒 AWS has announced general availability of the AWS for SAP MCP Server on Amazon Bedrock AgentCore, enabling AI agents to connect directly and securely to SAP ERP systems at scale. Built on Model Context Protocol (MCP) and SAP OData standards, the server supports CRUD access to sales orders, purchase orders, materials, and finance documents. The managed AgentCore Runtime provides session isolation, private connectivity, and dual-layer authentication with CloudWatch telemetry and CloudFormation templates for rapid, no-infrastructure deployment.

📊 AWS has launched the Attributed Revenue dashboard in AWS Partner Central, providing Partners a self-service view of monthly attributed revenue by product, AWS service, and billing period. The dashboard consolidates data from the three Partner Revenue Measurement methods — Resource Tagging, User Agent string, and AWS Marketplace Metering — so Partners can monitor consumption patterns and revenue trends. Partners with multiple AWS Marketplace seller accounts can connect subsidiary accounts to see aggregated revenue across all connected accounts, and the dashboard is available in all commercial regions for Partners migrated to AWS Partner Central.

🚀 Second-generation AWS Outposts racks are now supported in the AWS Asia Pacific (Seoul, Sydney) and Europe (Paris) Regions. Outposts racks bring AWS infrastructure, services, APIs, and tools into on‑premises data centers or colocation spaces to provide a consistent hybrid experience. Customers can order racks attached to these Regions to optimize latency, meet data residency requirements, run low-latency workloads locally, and manage applications from their home Region.

🔍 AWS has extended Elastic Beanstalk's AI-powered environment analysis to Windows Server platforms, enabling developers and operators to diagnose and resolve Windows-based environment issues more quickly. The feature collects recent events, instance health metrics, and logs from Windows instances and sends that telemetry to Amazon Bedrock for analysis. You can request an AI analysis from the Elastic Beanstalk console using the AI Analysis button or programmatically via the AWS CLI with the RequestEnvironmentInfo and RetrieveEnvironmentInfo operations. Results include step-by-step troubleshooting recommendations tailored to the environment's current state and are available in Regions where both services are offered.

🔒 Amazon Quick now supports adding co-owners to admin-managed Microsoft SharePoint Online and Google Drive knowledge bases and their data source connections. Owners receive full management capabilities — editing, syncing, sharing, and deleting — while Viewers have query-only access. The Owner co-owner option is restricted to admin-managed SharePoint and Google Drive; other knowledge base types support Viewer sharing only. Administrators can also share connections so teams can create knowledge bases from the same integration. The feature is available in all AWS Regions where Amazon Quick is offered.

🆕 Amazon Redshift now supports row-level UPDATE, DELETE, and MERGE operations on Apache Iceberg tables, allowing direct DML on both partitioned and unpartitioned Iceberg datasets. Customers can perform UPSERT workflows such as change data capture and slowly changing dimensions without moving data to external engines. Tables remain interoperable with other Iceberg-compatible engines and respect AWS Lake Formation permissions. The feature is available in all AWS Regions where Redshift is offered.

🔒 Amazon Quick now includes an ACL Permission Checker for knowledge bases with document-level ACLs, enabling administrators to verify whether a specific user can access a particular document without manually tracing permission inheritance. To use it, open a knowledge base with ACLs enabled, go to the Sync reports tab, choose View Access Details for any synced item, and enter the user's email in the Permission Checker to get an immediate result. The Access Details panel also lists all users and groups with access so administrators gain full visibility into applied permissions. The checker returns one of three outcomes: the user has access, the user does not have access, or no ACL was found for the document.

🔒 Amazon Quick now supports document-level access controls (ACLs) for Microsoft SharePoint knowledge bases, allowing organizations to preserve native SharePoint permissions when indexing content. Quick uses a dual approach—ACL replication for fast pre-retrieval filtering paired with real-time permission checks against SharePoint at query time—to avoid stale or incorrectly mapped access. Administrators can enable this in an admin-managed SharePoint knowledge base in the Quick console; the feature is available in all Regions where Quick is offered.

🔒 UNC6692 used persistent social engineering to lure victims via Microsoft Teams, delivering a staged payload that installed an AutoHotkey loader and a malicious Chromium extension (SNOWBELT) from attacker-controlled AWS S3. The intruders deployed a modular suite — SNOWBELT, SNOWGLAZE, and SNOWBASIN — to establish WebSocket tunnels, local HTTP backdoors, and stealthy proxying for lateral movement. The campaign combined credential theft, LSASS and NTDS extraction, and exfiltration to cloud services, highlighting the need to monitor browser extensions and cloud egress.