Kategorie: Amazon AWS

Amazon SageMaker HyperPod now supports G7e and r5d.16xlarge instances

Amazon SageMaker HyperPod now supports G7e and r5d.16xlarge instances. SageMaker HyperPod is a purpose-built infrastructure for developing, training, and deploying foundation models at scale. It provides a resilient and performant environment with built-in fault tolerance, automated cluster recovery, and optimized distributed training libraries, reducing the undifferentiated heavy lifting of managing large-scale AI/ML infrastructure. 
G7e instances are powered by NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs and deliver up to 2.3x better inference performance than G6e instances, allowing you to process more requests per second while reducing latency. With up to 768 GB of total GPU memory, G7e instances let you deploy larger language models or run multiple models on a single endpoint. You can use these instances for deploying LLMs, agentic AI, multimodal generative AI, and physical AI models. G7e instances are also well suited for cost-efficient single-node fine-tuning or training of NLP, computer vision, and smaller generative AI models, with up to 1.27x the TFLOPs and up to 4x the GPU-to-GPU bandwidth compared to G6e. In addition, HyperPod now supports r5d.16xlarge as well. The r5d.16xlarge instance provides 64 vCPUs, 512 GB of memory, and 5 x 600 GB NVMe SSD instance storage, powered by Intel Xeon Platinum 8000 series processors with a sustained all-core turbo frequency of up to 3.1 GHz. This instance is well suited for distributed training data preprocessing especially with frameworks such as Ray, large-scale feature engineering, and running memory-heavy orchestration services alongside GPU compute. G7e instances are available in US East (N. Virginia), US East (Ohio), Asia Pacific (Tokyo), and US West (Oregon) and r5d.16xlarge is available in all regions Amazon SageMaker HyperPod is available in. 
Quelle: aws.amazon.com

Amazon Connect now provides eight new metrics to measure and improve AI agent performance

Amazon Connect now provides eight new metrics to measure and improve AI agent performance, including goal success rate, faithfulness score, and tool selection accuracy. These metrics offer visibility into the quality of AI-driven customer interactions, enabling measurement and continuous improvement of AI agent outcomes. With this launch, you can monitor whether AI agents successfully resolved customer requests, assess faithfulness and detect contextual hallucinations. You can also evaluate tool selection and utilization accuracy, and capture customer feedback through thumbs up/down ratings when enabled.  You can access these new metrics through Amazon Connect’s AI Agent Performance dashboard, or through the GetMetricDataV2 API and zero-ETL data lake for custom reporting or integration with your existing analytics workflows.
This feature is available in all AWS Regions where Amazon Connect AI Agents is supported. For more information, see the Amazon Connect Administrator Guide. To learn more about Amazon Connect, an AI-native solution that turns every customer interaction into a moment worth remembering, visit the Amazon Connect website
Quelle: aws.amazon.com

Amazon Bedrock AgentCore Gateway and Identity support VPC egress

Amazon Bedrock AgentCore Gateway and Identity now provide secure and controlled egress traffic management for your applications, enabling seamless communication with resources in your Virtual Private Cloud (VPC). VPC egress for AgentCore Gateway targets and Identity credential providers are offered in both managed and self-managed configurations.
With VPC egress support, customers can now invoke private resources (e.g., EKS-hosted MCP servers) directly from their AgentCore Gateway. Managed VPC egress covers most customer use cases. For more complex networking setups, customers can configure their own VPC Lattice resources. AgentCore Identity VPC egress supports connectivity to Identity Providers (IdPs) running inside a customer’s VPC. This enables two key capabilities: validating inbound access tokens issued by your private IdP and fetching tokens from your IdP for outbound request authentication. Finally, this launch supports private DNS resolution for managed VPC egress resources across Gateway and Identity.
AgentCore Gateway and Identity are available in fourteen AWS Regions: US East (N. Virginia), US East (Ohio), US West (Oregon), Canada (Central), Asia Pacific (Mumbai), Asia Pacific (Seoul), Asia Pacific (Singapore), Asia Pacific (Sydney), Asia Pacific (Tokyo), Europe (Frankfurt), Europe (Ireland), Europe (London), Europe (Paris), and Europe (Stockholm).
Learn more about VPC egress capabilities through AgentCore Gateway documentation, and AgentCore Identity documentation. Get started with the AgentCore CLI.
Quelle: aws.amazon.com

Amazon Quick now integrates with Visier’s Vee agent for workforce intelligence

Amazon Quick now integrates with Vee, the AI assistant from Visier’s people analytics platform, through the model context protocol (MCP). HR business partners, finance managers, and operations leaders can now get governed access to live workforce intelligence from Visier directly within their Amazon Quick workspace without switching tools.
After setting up the connection in Quick using Visier’s remote MCP server, you can ask questions in natural language about headcount, attrition, tenure, and open requisitions and receive answers grounded in Visier’s governed workforce data model. Vee can also be invoked from automated Quick Flows to run recurring workforce reviews or draft documents. Quick intelligently routes relevant prompts to Vee and returns contextualized answers alongside enterprise knowledge – such as budgets, policies, and plans stored in Quick Spaces – so every answer reflects the full organizational picture.
The Visier integration with Amazon Quick is available in all AWS Regions where Amazon Quick is available.
To get started with Amazon Quick, visit the website. To learn more about the Visier integration, read the Visier integration guide, see the blog, and explore more integrations on the integrations page.
Quelle: aws.amazon.com

AWS Lambda Provisioned Mode for Kafka event source mappings (ESMs) now available in AWS Asia Pacific (Taipei) and AWS GovCloud (US) Regions

AWS Lambda now supports Provisioned Mode for event source mappings (ESMs) that subscribe to Apache Kafka event sources in the Asia Pacific (Taipei), AWS GovCloud (US-East), and AWS GovCloud (US-West) Regions. Provisioned Mode allows you to optimize the throughput of your Kafka ESM by provisioning event polling resources that remain ready to handle sudden spikes in traffic, helping you build highly responsive and scalable event-driven Kafka applications with stringent performance requirements.
Customers building streaming data applications often use Kafka as an event source for Lambda functions, relying on Lambda’s fully managed ESM to automatically scale polling resources in response to events. However, for event-driven Kafka applications that need to handle unpredictable bursts of traffic, lack of control over the throughput of ESM can lead to delays in your users’ experience. Provisioned Mode for Kafka ESM enables customers to fine-tune the throughput of their Amazon Managed Streaming for Apache Kafka (MSK) ESM or self-managed Kafka ESM by provisioning and auto-scaling between a minimum and maximum number of polling resources called event pollers. With this launch, this feature is now available in three additional regions.  
You can activate Provisioned Mode for MSK ESM or self-managed Kafka ESM by configuring a minimum and maximum number of event pollers in the ESM API, AWS Console, AWS CLI, AWS SDK, and AWS CloudFormation. You pay for the usage of event pollers, along a billing unit called Event Poller Unit (EPU). To learn more, read the Lambda ESM documentation and AWS Lambda pricing. 
Quelle: aws.amazon.com

AWS Client VPN now supports native AWS Transit Gateway integration

AWS Client VPN now supports native integration with AWS Transit Gateway, simplifying centralized remote access for your end users across multiple VPCs and on-premises, and providing end-to-end source IP visibility. AWS Transit Gateway interconnects your Amazon Virtual Private Clouds (VPCs) and on-premises networks, while AWS Client VPN enables secure remote access to AWS and on-premises resources connected through your AWS network. Previously, connecting Client VPN to multiple VPCs required provisioning and managing an intermediate VPC, adding operational complexity as you needed to manage additional resources. Moreover, client source IPs were translated through Source Network Address Translation (SNAT), making it difficult to identify which remote user generated specific traffic and complicating security audits. Native Transit Gateway attachment eliminates the need for an intermediate VPC, letting you provide centralized remote access to multiple VPCs and on-premises networks directly from your Client VPN endpoint. Additionally, the end-user source IP is now preserved end-to-end, so you can create authorization rules based on actual client IPs and trace traffic back to specific users, simplifying security, compliance, and troubleshooting workflows. Furthermore, Transit Gateway flow logs capture connection-level details tied to preserved source IPs for improved troubleshooting and compliance audits. This integration is available in all AWS Regions where AWS Client VPN is available. There are no additional charges for this native integration beyond standard pricing of AWS Client VPN and AWS Transit Gateway.
To learn more about Client VPN:

Visit the AWS Client VPN product page
Read the AWS Client VPN documentation

Quelle: aws.amazon.com

Amazon EC2 High Memory U7i instances now available in additional regions

Amazon EC2 High Memory U7i-8TB instances (u7i-8tb.112xlarge) are now available in AWS Europe (Stockholm, Zurich) regions, U7in-16TB instances (u7in-16tb.224xlarge) are now available in the AWS US East (Ohio) region, and U7in-24TB instances (u7in-24tb.224xlarge) are now available in the AWS Europe (Stockholm) region. U7i instances are part of the AWS 7th generation and are powered by custom fourth-generation Intel Xeon Scalable processors (Sapphire Rapids). U7i-8TB instances offer 8 TiB of DDR5 memory, U7in-16TB instances offer 16 TiB of DDR5 memory, and U7in-24TB instances offer 24 TiB of DDR5 memory, enabling customers to scale transaction processing throughput in a fast-growing data environment.
U7i-8TB instances deliver 448 vCPUs and support up to 100 Gbps of Amazon EBS bandwidth, 100 Gbps of network bandwidth, and ENA Express. Both U7in-16TB and U7in-24TB instances deliver 896 vCPUs and support up to 100 Gbps of Amazon EBS bandwidth for faster data loading and backups, 200 Gbps of network bandwidth, and ENA Express. U7i instances are ideal for customers running mission-critical in-memory databases like SAP HANA, Oracle, and SQL Server.
To learn more about U7i instances, visit the High Memory instances page.
Quelle: aws.amazon.com

Amazon SageMaker HyperPod now supports automatic Slurm topology management

Amazon SageMaker HyperPod now automatically selects and continuously maintains the optimal network topology configuration for Slurm clusters based on the GPU instance types in the cluster. Network topology directly impacts distributed training performance — when jobs are placed on nodes that are topologically close, GPU-to-GPU communication is faster, NCCL collective operations are more efficient, and training throughput improves. HyperPod dynamically adapts the topology as the cluster evolves through scaling operations and node replacements, so job placement remains optimized throughout the cluster lifecycle without requiring manual updates to topology files or Slurm reconfiguration. HyperPod inspects the instance types across all instance groups at cluster creation, identifies the networking and interconnect characteristics of each instance type, and automatically selects the best-fit topology model. HyperPod supports tree topology for instance types with hierarchical interconnects such as ml.p5.48xlarge, ml.p5e.48xlarge, and ml.p5en.48xlarge, and block topology for instance types with uniform high-bandwidth connectivity such as ml.p6e-gb200.NVL72. For clusters with mixed instance types, HyperPod selects a compatible topology that works across all nodes. As the cluster changes through scale-up, scale-down, or node replacement events, HyperPod automatically updates the topology configuration without manual intervention, so the topology always reflects the actual state of the cluster.
To get started, create a SageMaker HyperPod Slurm cluster with supported GPU instance types. Topology-aware scheduling is enabled by default and requires no configuration.
This feature is available in all AWS Regions where Amazon SageMaker HyperPod is supported. To learn more about topology-aware scheduling, visit the Amazon SageMaker HyperPod documentation
Quelle: aws.amazon.com

AWS Parallel Computing Service now supports Slurm 25.11

AWS Parallel Computing Service (AWS PCS) now supports Slurm version 25.11, with support for a Prometheus-compatible OpenMetrics endpoint, and introduces new log types including scheduler audit logs. This release of Slurm 25.11 introduces expedited re-queue, which can automatically reschedule jobs affected by node issues at the highest priority to help your workloads recover faster. You can enable a new OpenMetrics endpoint for real-time visibility into jobs, nodes, and scheduling using your existing monitoring tools. AWS PCS can now also send Slurm database daemon (slurmdbd) and REST API daemon (slurmrestd) logs to Amazon CloudWatch Logs, Amazon S3, or Amazon Data Firehose, helping diagnose accounting issues and debug API integrations. Scheduler audit logs, previously included in operational logs, are now delivered as a dedicated log type, providing independent control over ingestion and storage costs. AWS PCS is a managed service that makes it easier for you to run and scale your high performance computing (HPC) workloads and build scientific and engineering models on AWS using Slurm. You can use AWS PCS to build complete, elastic environments that integrate compute, storage, networking, and visualization tools. AWS PCS simplifies cluster operations with managed updates and built-in observability features, helping to remove the burden of maintenance. You can work in a familiar environment, focusing on your research and innovation instead of worrying about infrastructure. These features are available in all AWS Regions where AWS PCS is available. Standard charges apply for log delivery destinations. To learn more about AWS PCS, refer to the service documentation.
Quelle: aws.amazon.com

Amazon Athena simplifies federated queries with managed connectors

Amazon Athena now offers managed connectors for 12 data sources, including Amazon DynamoDB, PostgreSQL, MySQL, and Snowflake. Managed connectors are AWS Glue Data Catalog federated connectors that Athena creates and manages on your behalf, so you can query data outside Amazon S3 without deploying or maintaining connector resources in your AWS account. With Athena, you can interactively query relational, non-relational, object, and custom data sources without moving or duplicating data. To get started with managed connectors, you create a connection for your data source in Athena. Athena automatically sets up and manages connector resources on your behalf, registering the data source as a federated catalog in AWS Glue Data Catalog. You can then query the data source alongside your Amazon S3 data and optionally set up fine-grained access controls through AWS Lake Formation. Federated queries with managed connectors are available in all AWS Regions where Athena is available, except the AWS GovCloud (US) Regions and the China Regions. To learn more, visit Use Amazon Athena Federated Query in the Athena User Guide.
Quelle: aws.amazon.com