All news with #aws tag

🎙️ Amazon SageMaker AI Inference now supports bidirectional streaming, enabling real-time speech-to-text transcription that returns partial transcripts while audio is still being captured. Using the new Bidirectional Stream API, clients open an HTTP/2 connection to the SageMaker AI runtime, which automatically creates a WebSocket to your model container so audio frames and interim transcripts flow continuously. Any container that implements a WebSocket handler per the SageMaker AI contract works out of the box, allowing real-time models such as Deepgram to run without modification. The feature eliminates weeks or months of custom streaming infrastructure work so teams can focus on model accuracy, latency tuning, and agent behavior.

⚡ Amazon SageMaker AI now supports EAGLE (Extrapolation Algorithm for Greater Language-model Efficiency) speculative decoding to boost large language model inference throughput by up to 2.5x. The capability enables models to predict and validate multiple tokens in parallel rather than one at a time, preserving output quality while reducing latency. SageMaker automatically selects between EAGLE 2 and EAGLE 3 depending on model architecture and provides built‑in optimization jobs using curated or customer datasets. Optimized models can be deployed through existing SageMaker inference workflows without infrastructure changes, and the feature is available in select AWS Regions.

🔧 AWS IoT Core now supports dynamic retrieval of thing registry data directly within IoT rules using the new get_registry_data() inline rule function. You can access device attributes, device type, and group membership to filter, enrich, and route messages — for example, routing lifecycle events or enriching gateway messages with sensor thresholds. The feature is available in all AWS regions where AWS IoT Core is present and can be used immediately via the IoT developer guide and API.

🔒 Amazon CloudFront now supports mutual TLS (mTLS), enabling both server and client authentication with X.509 certificates at AWS edge locations. Customers can require trusted client certificates to access distributions, reducing unauthorized access to APIs and applications. Typical uses include secure B2B API integrations and IoT device authentication. mTLS is available at no additional cost and configurable via Console, CLI, SDK, CDK, and CloudFormation.

🔁 Amazon EC2 now offers interruptible On‑Demand Capacity Reservations (ODCRs), allowing reservation owners to temporarily make unused reserved capacity available to other internal workloads while retaining the ability to reclaim it. Consumers using interruptible ODCRs receive an interruption notice to allow graceful shutdown or checkpointing. The capability targets flexible, fault‑tolerant tasks such as batch processing, data analysis, and machine learning training, is available at no additional cost, and CloudFormation support is coming soon.

🔍 Amazon OpenSearch Service now makes Piped Processing Language (PPL) and natural language the default experience in the OpenSearch UI Observability workspace. The release introduces 35+ new commands for deep analysis, faceted exploration, and natural-language querying, and integrates OpenTelemetry ingestion pipelines to simplify onboarding. Users can run enterprise-grade queries, correlate events, and move directly from query to visualization to reduce mean time to detect and resolve issues.

⚠️ Researchers disclosed five vulnerabilities in Fluent Bit, the open-source telemetry agent, that can be chained to bypass authentication, write or overwrite files, execute code, corrupt logs, and cause denial-of-service conditions. CERT/CC noted many issues require network access, and fixes were released in Fluent Bit 4.1.1 and 4.0.12 with AWS participating in coordinated disclosure. Operators are urged to update immediately and apply mitigations such as avoiding dynamic tags, mounting configs read-only, and running the agent as a non-root user.

🎥 AWS Elemental MediaTailor now supports HLS Interstitials for live streams, enabling broadcasters and streaming services to insert interstitial ads using the HLS Interstitials specification. MediaTailor generates necessary metadata tags (EXT-X-DATERANGE with X-ASSET-LIST) for compatible players such as HLS.js, Shaka, Bitmovin and Apple devices, removing the need for custom client stitching. The capability integrates with MediaTailor's server-side ad insertion for frame-accurate, buffer-free transitions and preserves server-side beaconing for measurement. It supports pre-roll and mid-roll, can be toggled per playback session via a manifest query parameter, and is available in all Regions where MediaTailor operates on a pay-as-you-go basis.

🧩 Amazon Connect flow modules now support custom inputs, outputs and branching logic, plus advanced versioning and alias management. You can define flexible parameters for reusable modules—for example, an authentication module that accepts a phone number and PIN and returns customer name and authentication status with branches like authenticated or not authenticated. Immutable version snapshots and alias mapping let teams update module implementations while flows referencing an alias automatically use the new version, simplifying maintenance and reuse.

💬 AWS announced general availability of Amazon Quick Suite Embedded Chat, a ready-made conversational AI you can embed into applications via one-click embedding or API-based iframes. The agent unifies structured data and unstructured knowledge in a single conversation so users can reference KPIs, pull file details, check customer feedback, and trigger actions without leaving the app. Connectors include SharePoint, websites, Slack, and Jira, and enterprises retain control over data access and action scopes. Embedded Chat is available in select Regions with no additional charge beyond existing Quick Suite pricing.

🔗 AWS announces general availability of AWS Glue catalog federation for remote Apache Iceberg catalogs. The feature enables analytics engines to query Iceberg tables stored in Amazon S3 and cataloged remotely without moving or copying data, with real-time metadata synchronization to the AWS Glue Data Catalog. It leverages AWS Lake Formation for fine-grained access controls and supports the Iceberg REST specifications; federation is available in the Lake Formation console and via SDKs/APIs.

🔐 Amazon Redshift now supports federated permissions to centralize and enforce data access policies across multiple Redshift warehouses, reducing governance overhead for multi-warehouse deployments. Registered warehouses are auto-mounted account-wide and can be queried using existing workforce identities via AWS IAM Identity Center or IAM roles. Row-level, column-level, and masking controls are applied automatically, ensuring consistent fine-grained access control regardless of query location.

🔁 AWS Lambda now offers enhanced error handling for Amazon Managed Streaming for Apache Kafka (MSK) and self-managed Kafka event sources, enabling developers to define custom retry configurations and route failed messages to a Kafka topic as an on-failure destination. The update extends Kafka event source mapping (ESM) in Provisioned mode to support retry limits, time-bound retry windows, automatic discards of exceeded records, and per-message failure reporting to optimize retries. Configure these settings via the ESM API, AWS Console, or AWS CLI.

🔁 Amazon has expanded availability of Amazon MSK Replicator, allowing customers to replicate streaming data across Amazon MSK clusters in five more AWS Regions: Asia Pacific (Thailand), Mexico (Central), Asia Pacific (Taipei), Canada West (Calgary), and Europe (Spain). MSK Replicator offers automatic asynchronous replication, auto-scales underlying resources on demand, and replicates Kafka metadata such as topic configurations, ACLs, and consumer group offsets. You can configure replication from the Amazon MSK console or the AWS CLI.

🚀 Amazon EC2 High Memory U7i instances (u7i-6tb.112xlarge) are now available in Asia Pacific (Jakarta), offering 6TB of DDR5 memory and 448 vCPUs powered by custom Intel Sapphire Rapids processors. The instances support up to 100Gbps for EBS and networking and include ENA Express for lower-latency, consistent network performance. They are positioned for mission-critical in-memory databases such as SAP HANA, Oracle, and SQL Server, enabling higher transaction throughput and faster data loading and backups.

🔐 A Trend Micro report warns that ransomware groups are shifting from on-premises targets to cloud object storage, particularly AWS S3, by abusing integrated encryption and key management. Attackers probe configurations from AWS-managed KMS keys to customer-provided and external key stores to encrypt or irreversibly lock data. The report urges hardening S3 settings, enforcing least privilege, enabling versioning and Object Lock, and isolating backups.

🌐 Amazon CloudFront now supports bringing your own IP addresses (BYOIP) for Anycast Static IPs through VPC IP Address Manager (IPAM). Network teams can register and manage public IPv4 address pools in IPAM and assign dedicated Anycast Static IP lists to CloudFront distributions, preserving existing allow-lists and avoiding changes to application address space. The capability simplifies IP address management across AWS's global edge network and improves partner reachability and security. It is available in all commercial AWS Regions except AWS GovCloud (US) and the China regions.

🚀 Amazon SageMaker HyperPod now supports NVIDIA Multi-Instance GPU (MIG), enabling administrators to partition a single GPU into multiple isolated devices to run simultaneous small generative AI tasks. Administrators can use an easy console configuration or a custom setup for fine-grained hardware isolation, allocate compute quotas across teams, and monitor real-time performance per partition via a utilization dashboard. Available on HyperPod clusters using the EKS orchestrator in multiple AWS Regions, this capability reduces wait times by letting data scientists run lightweight inference and interactive notebooks in parallel without consuming full GPU capacity.

⚡ Amazon SageMaker HyperPod now supports Spot Instances, enabling customers to reduce GPU compute costs by up to 90% compared with on-demand instances. The integration is available on HyperPod EKS clusters and works with Karpenter for intelligent autoscaling, automatic Spot capacity discovery, and interruption handling. You can enable Spot when creating instance groups via the CreateCluster API or the AWS Console, and the feature supports all HyperPod instance types across available regions.

🔒 Amazon Aurora PostgreSQL-Compatible Edition now supports dynamic data masking using the new pg_columnmask extension, enabling column-level protection at query time. The extension complements PostgreSQL row-level security and column grants by letting administrators define SQL-based masking policies that alter how data appears to users without changing stored values. Policies can use built-in or user-defined functions to hide, partially mask, or transform data, and multiple policies can be applied with weighted precedence. pg_columnmask protects results across WHERE, JOIN, ORDER BY, and GROUP BY clauses and is available for Aurora PostgreSQL 16.10+ and 17.6+ in all regions.