AWS News - 2026-08-25

Amazon SageMaker HyperPod now enhances support for Ray with built-in observability, resilient training, accelerated inference and managed development environments. Ray is a popular open-source framework for scaling AI workloads on a unified compute layer, from data processing and distributed training to reinforcement learning and model serving. Running Ray on Kubernetes at production scale can be an operational burden: job hangs, low GPU utilization from static team allocations, and multi-step observability setup. Also, lack of interactive development environment means every code change needs another job submission and familiarity with kubectl.

HyperPod now brings easier development, resilient training, and accelerated inference to Ray. Data scientists create, edit, monitor, and delete Ray clusters from a web-based interface in Amazon SageMaker Studio, then attach JupyterLab, Code Editor, or a local IDE to a running Ray cluster and iterate interactively against cluster-scale compute. A multi-node Ray cluster behaves like a local development environment, so you test each change immediately, without waiting for a new job to queue and start. For Observability, HyperPod provisions Grafana dashboards with metrics in Amazon Managed Service for Prometheus and allows one-click access to the Ray Dashboard through a secure browser link, giving you visibility into your workloads from the first run. For training at scale, HyperPod node auto recovery and hung job detection handle GPU faults, job hangs, loss spikes, and degraded throughput. Tiered checkpointing restores state from cluster memory to maximize goodput, and task governance improves compute utilization through quotas, priorities, and preemption. Together, these keep your long training runs progressing through failures and maximize the useful work done per GPU-hour. For inference with Ray Serve, a tiered KV cache reuses cached prefixes to reduce time to first token, and you can deploy Amazon SageMaker JumpStart models directly.

Open-source Ray code runs unchanged and you can either adopt the purpose-built experience in SageMaker Studio or take individual capabilities to integrate into your own ML platform.

Ray support is available for HyperPod clusters orchestrated by Amazon EKS, in AWS Regions where SageMaker HyperPod is supported. To learn more, see the SageMaker HyperPod documentation, and explore the interactive demo.

Amazon MSK Replicator now supports OAuth 2.0 (SASL/OAUTHBEARER) authentication for data replication from external Apache Kafka clusters - including on-premises, self-managed on AWS, or other cloud providers - to Amazon MSK Provisioned clusters. With this capability, external Apache Kafka clusters configured with OAuth/OIDC authentication can now use MSK Replicator to migrate workloads to Amazon MSK, support disaster recovery by using MSK-based clusters as a failover or backup target and enable data distribution across hybrid and multi-cloud environments.

MSK Replicator is a feature of Amazon MSK that automates data replication between Kafka clusters, eliminating the need to manage custom replication infrastructure or configure open-source tools. Previously, MSK Replicator supported SASL/SCRAM and mTLS authentication for connecting to external Apache Kafka clusters. With this launch, you can now also use OAuth 2.0 authentication with MSK Replicator to replicate data from external Kafka clusters to Amazon MSK. Unlike self-managed replication tools, MSK Replicator lets you retain your original Kafka topic names during replication while automatically avoiding infinite replication loops. It also synchronizes consumer group offsets bidirectionally, enabling you to move producers and consumers across clusters independently, in any order, without coordination constraints or the risk of data loss.

This new capability is supported in all AWS Regions where MSK Replicator is available. Visit the MSK Replicator documentation, product page, pricing page, and this AWS blog post to learn more.

AWS Lambda functions now support full Identity and Access Management (IAM) resource-based policies, enabling platform admins and security teams to define granular access permissions using the full capabilities of AWS IAM. With full IAM resource-based policies, you can define permissions for multiple principals and actions in a single policy document and leverage the full range of IAM condition keys.

Previously, Lambda functions required customers to add permissions individually per principal. This provided limited flexibility for platform admins and security teams who want to manage permissions at scale. Now, Lambda functions support full IAM resource-based policies, including the full range of IAM condition keys. This provides a broader range of policy capabilities and streamlines policy management for teams operating multi-account architectures or managing multiple resources. For example, you can now use IAM condition keys to restrict access based on source IP or principal tag, and platform teams can now allow multiple services to invoke a function by using a single policy, rather than maintaining multiple statements to add permissions.

You can update resource-based policies in one step using the JSON editor in the AWS Lambda console, AWS CLI, AWS SDK, or infrastructure as code tools such as AWS CloudFormation and AWS SAM. To learn more, explore the Lambda resource-based policy examples in the AWS Lambda Developer Guide. 

Full IAM resource-based policies are available in all AWS commerical Regions at no additional charge.

Amazon Relational Database Service (Amazon RDS) for Oracle now supports the Oracle July 2026 Release Update (RU) for Oracle Database versions 19c, 21c and 26ai. We recommend upgrading to the July 2026 RU as it includes security updates for Oracle database products.

Starting with July 2026 releases, the naming format for an RU in Oracle Database 19c changes to <version>.ru-<YYYY-MM>.mrp-<YYYY-MM>.r<N>. For example, the RDS for Oracle July 2026 quarterly RU for Oracle Database 19c is named 19.0.0.0.ru-2026-07.mrp-2026-07.r1. When Oracle releases a monthly Critical Security Patch Update (CSPU) for Oracle Database 19c and 26ai, Amazon RDS will make it available as an MRP version that bundles the CSPU with additional Oracle-recommended fixes. Oracle Database 21c RUs retain their existing naming format. For details, see Release updates and monthly recommended patches. 

You can apply the July 2026 RU from the Amazon RDS Management Console, or by using the AWS SDK or CLI. To automatically apply updates to your database instance during your maintenance window, enable Automatic Minor Version Upgrade.

You can also use AWS Organizations upgrade rollout policy to stagger automatic minor version upgrades for your Amazon RDS database instances. This feature allows you to automatically apply updates to non-production environments, validate the updates, and then automatically apply the same update to production environments. For additional details about using AWS Organizations upgrade rollout policy for automatic minor version upgrades, refer to Amazon RDS for Oracle documentation. 

Today, AWS Lambda introduces managed runtimes in public preview, starting with Node.js 26 and Python 3.15. This gives customers, partners and upstream language communities the opportunity to try upcoming managed runtimes before their general availability (GA) release and to share their feedback with the Lambda team.

Previously, Lambda managed runtimes were launched directly as GA releases. Since the runtimes could immediately be used by production workloads, the Lambda team couldn’t make changes post-GA that might impact existing functions. Managed runtimes in public preview give you a chance to test your workloads and report issues before GA—while breaking changes are still possible, so we can act on your feedback. Third-party observability providers, infrastructure-as-code tools and deployment frameworks also get time to validate compatibility before GA. Because public preview managed runtimes may receive breaking changes during the preview period, they are not covered by the Lambda SLA or AWS Technical Support plans and should not be used for production workloads. Managed runtimes in public preview use the same runtime identifier as the eventual GA runtime, so your functions graduate automatically when the runtime reaches GA with no action required.

To get started, select "Node.js 26 (Preview)" or "Python 3.15 (Preview)" from the runtime dropdown in the Lambda console, or specify nodejs26.x or python3.15 as the runtime identifier in the AWS CLI, CloudFormation, AWS SAM, or AWS CDK.

Node.js 26 and Python 3.15 managed runtimes in public preview are available in all AWS commercial Regions, AWS GovCloud (US) Regions, and China Regions. There is no additional cost—functions using preview runtimes are billed at standard Lambda rates.  To learn more, see our blog post. To share feedback, visit the dedicated GitHub issues for Node.js 26 and Python 3.15.

Starting today, Amazon EC2 M8i and M8i-flex instances are now available in Canada West (Calgary) region. These instances are powered by custom Intel Xeon 6 processors, available only on AWS, delivering the highest performance and fastest memory bandwidth among comparable Intel processors in the cloud. The M8i and M8i-flex instances offer up to 15% better price-performance, and 2.5x more memory bandwidth compared to previous generation Intel-based instances. They deliver up to 20% better performance than M7i and M7i-flex instances, with even higher gains for specific workloads. The M8i and M8i-flex instances are up to 30% faster for PostgreSQL databases, up to 60% faster for NGINX web applications, and up to 40% faster for AI deep learning recommendation models compared to M7i and M7i-flex instances. 
 
M8i-flex instances are the easiest way to get price performance benefits for a majority of general-purpose workloads like web and application servers, microservices, small and medium data stores, virtual desktops, and enterprise applications. They offer the most common sizes, from large to 16xlarge, and are a great first choice for applications that don't fully utilize all compute resources. 
 
M8i instances are a great choice for all general-purpose workloads, especially for workloads that need the largest instance sizes or continuous high CPU usage. The SAP-certified M8i instances offer 13 sizes including 2 bare metal sizes and the new 96xlarge size for the largest applications. 
 
To get started, sign in to the AWS Management Console. For more information about the new instances, visit the M8i and M8i-flex instance page or visit the AWS News blog.