AWS News - 2026-10-28

Amazon ElastiCache for Valkey now lets you enable or disable durability on your existing node-based clusters without creating a new cluster. Previously, you couldn't change the durability configuration of an existing cluster in place. This made it harder to move from caching to persistent data store use cases, or to adapt as your application requirements change. Now you can change durability at any point in your application's lifecycle.

You can choose between two durability options: synchronous and asynchronous writes. Synchronous writes persist data across at least two AZs before responding to the client, designed for zero data loss at single-digit millisecond write latency. Asynchronous writes persist data after responding to the client, maintaining microsecond write latency at no additional cost. However, up to 10 seconds of uncommitted data could be lost in the rare event of a failure. Both options maintain microsecond read latency. You can now use ElastiCache for a broader set of use cases beyond caching where data loss is unacceptable such as AI agent long-term memory, AI agent workflow state, knowledge bases for RAG applications, payment tokenization, and real-time inventory management.

This feature is available in all AWS Regions where ElastiCache for Valkey 9.0 and later is available. To get started, use the AWS Management Console, AWS SDK, or AWS CLI. For pricing, see the  Amazon ElastiCache pricing page . To learn more, see  Durability  in the ElastiCache documentation.

Amazon ElastiCache for Valkey data tiering is now available on Graviton4-based R8gd nodes in the AWS Europe (Spain) Region, and on Graviton3-based R7gd nodes in the Europe (Spain) and Asia Pacific (Hyderabad) Regions. Data tiering is a lower-cost way to scale your Valkey clusters to hundreds of terabytes of capacity. It uses solid state drives (SSDs) on each node alongside memory. When memory fills up, ElastiCache automatically moves the least recently used items to SSD. When your application reads an item stored on SSD, ElastiCache moves it back into memory before serving the request. Data tiering is a good fit for workloads that regularly access up to 20% of their dataset, and for applications that can tolerate some extra latency when they read data from SSD.

R8gd nodes are powered by AWS Graviton4 processors. They deliver up to 47% higher throughput, up to 43% lower P99 latency, and up to 31% better price-performance than Graviton3-based nodes of the same size. R7gd nodes are powered by AWS Graviton3 processors. They deliver up to 28% more throughput (read and write operations per second), up to 21% better P99 latency, and up to 25% more network bandwidth than Graviton2-based nodes. Actual performance varies by node size and workload.

You can use data tiering on R8gd and R7gd nodes with Valkey 7.2 and later. To get started, create a new cluster or upgrade an existing one with the AWS Management Console, the AWS CLI, or the AWS SDKs. For pricing, see the  Amazon ElastiCache pricing page . To learn more, see  Data tiering  and  Supported node types  in the ElastiCache documentation.

Amazon Connect Customer lets agents preview attachments directly in the agent workspace, reducing contact handling time and limiting the need to download sensitive customer information. Agents can preview PDFs, images, and text files across email, chat, and cases, navigate between attachments, and view multi-page documents without leaving the conversation.

Amazon Connect Customer attachment previews is available in the US East (N. Virginia), US West (Oregon), Africa (Cape Town), Asia Pacific (Seoul), Asia Pacific (Singapore), Asia Pacific (Sydney), Asia Pacific (Tokyo), Canada (Central), Europe (Frankfurt), and Europe (London) regions.

To learn more and get started, please refer to the help documentation or visit the Amazon Connect Customer website.

Amazon SageMaker AI MLflow Apps now support MLflow 3.16, bringing a redesigned tracing experience and faster experiment tracking to your generative AI and machine learning workflows.

MLflow 3.16 improves debugging and analysis of generative AI applications and agents. The redesigned trace explorer is now the default, with adjustable row height, reorderable columns, custom columns from any trace tag or metadata field. This release also accelerates metric charts for experiments with large metric histories.

New MLflow Apps are created on MLflow 3.16 by default. Existing MLflow Apps will be upgraded to MLflow 3.16 during an upcoming maintenance window, with no action required.

MLflow 3.16 is available in all AWS Regions where MLflow Apps are available. To learn more, visit the SageMaker AI MLflow developer guide.

Amazon Elastic Container Service (Amazon ECS) now supports GPU sharing with NVIDIA Multi-Process Service (MPS), enabling you to run multiple Amazon ECS tasks concurrently on a single NVIDIA GPU. GPU sharing helps you increase GPU utilization and reduce infrastructure costs for workloads that use only part of a GPU's capacity, such as small-model inference, AV/ADAS simulations, embedding generation, and multi-model serving.

With GPU sharing, you can pack more workloads onto each GPU instance without changing your application code. You define the GPU memory each container needs and can optionally set a maximum share of GPU compute, and Amazon ECS places tasks on GPUs with available capacity as your services scale and deploy. GPU sharing is supported on Amazon ECS Managed Instances and Amazon ECS on Amazon EC2. On Amazon ECS Managed Instances, Amazon ECS also monitors the MPS control daemon that enables sharing on each instance. If the daemon stops or becomes unhealthy, Amazon ECS automatically replaces the instance and moves your tasks to healthy capacity, so your GPU workloads keep running without manual intervention.

To get started, use the AWS Management Console, AWS CLI, AWS SDKs, AWS CloudFormation, or other infrastructure-as-code tools to create a capacity provider with NVIDIA GPU instances, set the GPU sharing strategy and memory in the container definition of your task definition, and create a service on that capacity provider. This capability is available in all AWS commercial and AWS GovCloud (US) Regions. To learn more, visit the Amazon ECS GPU sharing documentation.

AWS Backup restore testing now supports Amazon Redshift. Restore testing helps you automate and periodically run restore tests of your backups, so you can confirm that your Amazon Redshift backups restore successfully and measure how long the restores take.

You can now include Amazon Redshift in your restore testing plans alongside other supported AWS resources across storage, compute, and database services. Restore testing plans automatically select recovery points, run restores on a schedule, and measure restore job completion time against your recovery time objectives (RTO). This helps you evaluate recovery readiness and meet regulatory and compliance requirements for disaster recovery and business continuity.

To get started, create a restore testing plan and assign your Amazon Redshift resources by Amazon Resource Name (ARN) or tag. Choose whether each test restores the latest or a random eligible recovery point. Optionally, use restore testing validation to confirm the restored resource before AWS Backup deletes the test resources.

This support is available in all AWS Regions where both restore testing and Amazon Redshift are supported by AWS Backup. You can use the AWS Backup console, AWS Command Line Interface (CLI), or AWS SDKs. For a complete list of supported Regions and features, visit the AWS Backup documentation. To learn more about restore testing, visit the product page and pricing page.

Today, AWS Marketplace and Amazon Business announced the unified storefront. Organizations can now give their teams one curated catalog that puts software, data, and AI tools from AWS Marketplace alongside physical goods and supplies from Amazon Business. Many enterprises already buy from both, but they manage two catalogs with duplicated approval rules and separate reporting. The unified storefront can now give procurement teams one front door for both.

Built on AWS Marketplace Storefront, the unified storefront lets administrators connect their organization's Amazon Business account and import Amazon Business products into their branded storefront. Administrators set governance rules once, and they apply across both catalogs. Buyers can browse AWS Marketplace listings and Amazon Business products in one place. Purchases of AWS Marketplace listings complete directly in the storefront and appear on the customer's AWS invoice. When a buyer selects an Amazon Business product, single sign-on takes them to an authenticated Amazon Business cart, so they don't need a second login. Each storefront connects only to the accounts its administrator sets up.

The Amazon Business connector is available in all AWS Regions where AWS Marketplace operates to AWS Marketplace Storefront customers who have access to both the AWS Marketplace and Amazon Business APIs.

To learn more, visit the AWS Marketplace Storefront product page and the AWS Marketplace Storefront User Guide.

AWS Agent Registry, available through Amazon Bedrock AgentCore, is now available in four additional AWS Regions: Europe (London) eu-west-2, Europe (Frankfurt) eu-central-1, US East (Ohio) us-east-2, and Canada (Central) ca-central-1.

AWS Agent Registry is a fully managed discovery service that provides a centralized catalog for organizing, curating, and discovering resources across your organization. With AWS Agent Registry, you can publish MCP servers, tools, agents, agent skills, and custom resources into a searchable registry, control access through an approval workflow, and enable both human users and AI agents to discover the right tools and agents through hybrid search, catalog browsing, and a native MCP endpoint.

AWS Agent Registry is now available in nine AWS Regions. For a complete list of Region availability, see the AWS Regional Services List.

To get started, see the AWS Agent Registry documentation: https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/registry.html

北米の大手電力会社が Amazon Bedrock AgentCore 上で運用しているマルチエージェント AI アーキテクチャを紹介します。このアーキテクチャにより、オペレーターがシフトごとに何十回も、複数のシステムをまたいで手作業で行っていた調査が不要になり、アラームのトリアージにかかる時間が数時間から数分に短縮されています。業界の課題、マルチエージェントの設計、GitHub 上のリファレンス実装について説明します。

コンタクトセンターへの生成 AI 導入は、すでに「試してみる」段階を越えつつあります。導入から数年が経ったコンタクトセンターでよく伺うのは、FAQ の回答精度が思ったほど伸びないという悩みです。窓口を増やすたびに開発が必要になる、利用量が増えたときのコストが読めない、といった二周目の課題もここに重なります。 株式会社ドコモSMTBネット銀行（以下、ドコモSMTBネット銀行）は、生成 AI による自動応対をすでに本番で運用していました。その基盤を Amazon Connect Customer AI agents に移し、2026 年 9 月から電話とチャットの両チャネルで応対を開始しました。本ブログでは、なぜ基盤を移したのか、そして本番導入に向けて何を作り込んだのかをご紹介します。

11 月 1 日から Kiro の無料トライアルが変わります。初めて Pro にアップグレードすると、その月の残りは無料になり、12 か月間使える 500 トライアルクレジットも付与されます。プラン料金とクレジットは秒単位で日割り計算されます。上位プランで開始した場合は、日割りの Pro 料金との差額のみを支払います。本記事では、変更内容、計算方法、15 日にアップグレードした場合の各プランの金額とクレジットを紹介します。

この記事は 2026 年 8 月 27 日に公開された Building audio provenance f […]

Kiro Crew チームが、週 1,000 件のプルリクエストをマージしながら品質を保つ方法を紹介します。人間がすべての行を読まなくても確信を得られるよう、レビューの重大度分け、役割別の AI レビュアー、80 件以上のチェックを 1 つにまとめる PR Readiness 判定、prepare-pr スキルによる自動修正ループ、マージ後のセキュリティ対策へと段階的に発展させた過程を解説します。過剰なエンジニアリングへの対処法も紹介します。

The most important work I’ve done in cybersecurity would make for terrible television. There were no countdown clocks and no frantic typing, just a lot of careful reading, a lot of hard conversations, and the occasional quiet moment where everything finally clicked. That’s the part I want to show you this week, because the three […]

We continue to expand the scope of our assurance programs at Amazon Web Services (AWS) and are pleased to announce the successful completion of our first ever Cloud High (PBHH) security assessment with 172 assessed services and features. Completion of this assessment, effective September 18, 2026, makes AWS the first cloud service provider (CSP) in […]

November 9, 2026: We added a link to the French translation of this post and made other minor edits for clarity. French translation We continue to expand the scope of our assurance programs at Amazon Web Services (AWS) and are pleased to announce the successful completion of our first ever Canadian Centre for Cyber Security […]

Your workloads reach most destinations by domain name, but a firewall can only act on what it sees in the traffic. For HTTP and HTTPS, AWS Network Firewall domain rules match the name that the client puts in the request, and the client chooses that name. The connections for protocols such as SSH and SSH […]

Bulletin ID: 2026-152-AWS
Scope: AWS
Content Type: Important (requires attention)
Publication Date: 10/27/2026 11:00 AM PDT

Description:

The AWS API MCP Server is an open source Model Context Protocol (MCP) server that enables AI assistants to interact with AWS services and resources through AWS CLI commands. It can be configured to serve the streamable-http transport with OAuth authentication, validating a bearer token against an identity provider nominated by the operator. We identified CVE-2026-115413 in versions >=1.3.0 and <1.6.0, the server validates a token's signature, issuer and expiry, but not the claim identifying the application the token was issued for.

Impacted versions: >=1.3.0 AND <1.6.0

Please refer to the article below for the most up-to-date and complete information related to this AWS Security Bulletin.

Read this article

Anchorage Digital and Turnkey built verifiable infrastructure on AWS Nitro Enclaves so crypto wallets can cryptographically verify which code parsed a transaction. This post covers the architecture, the attestation-gated access pattern, and how it extends to any high-trust workflow beyond blockchain.

Manta Marine Technologies closes the gap between seemingly identical ships by training machine learning models on each vessel's own operational history. Learn how Manta delivers vessel-specific voyage optimization on AWS to help shipowners reduce fuel consumption and improve voyage efficiency.

See how to use Kiro, an agentic AI, to automate converting raw training material into compliant SCORM packages that are ready to upload to a learning management system (LMS).

In this post, we show how Exowatt used Amazon SageMaker AI serverless model customization to build a repeatable AI pipeline that produces task-specific fine-tuned models for two deployment targets, edge and cloud, achieving higher accuracy than frontier models at up to 49x lower cost.

Property Finder found that 96.8% of their vision-AI evaluation labels were inherited from the incumbent model, hiding its true accuracy. See how they built unbiased ground truth, applied evolutionary prompt optimization, and selected a more accurate, cost-effective model on Amazon Bedrock, cutting cost by 86% across 11 image-quality attributes.

Seekable OCI (SOCI) indexing for Amazon SageMaker AI training and processing jobs takes full container image download off the critical path. SageMaker AI lazy-loads your image to start containers in seconds instead of minutes, with no changes to your job definition. See the benchmarks and how to adopt it.

AWS has been named a Leader in the 2026 Gartner Magic Quadrant for Contact Center as a Service (CCaaS) for AWS Connect Customer. This recognition marks an important milestone for our team, and it reflects years of work pushing on what's possible in customer experience. 

Learn how an AWS Lambda function and an Amazon Bedrock AgentCore agent investigate a folder of related documents together through a shared Amazon S3 Files file system, using ordinary file operations instead of object download and upload code.