Publicado el — Deja un comentario

Ayudar a familias y educadores a apoyar experiencias más seguras en Windows

Ayudar a familias y educadores a apoyar experiencias más seguras en Windows

Un adolescente y una persona adulta usan una PC en casa

Por: Rob Mauceri, ingeniero distinguido, Seguridad Digital de Windows.

Nuestros niños y adolescentes crecen en un mundo digital donde las aplicaciones, juegos y herramientas impulsadas por IA forman parte de cómo aprenden, juegan, crean y se conectan. Para padres y educadores, eso aporta tanto oportunidad como responsabilidad: las experiencias digitales deben ayudar a los jóvenes a explorar con confianza mientras apoyan su bienestar, establecen hábitos saludables, respetan su privacidad y ofrecen protecciones adecuadas a su edad.

En Microsoft, creemos que la seguridad debería ser más fácil de entender para las familias y más fácil para que los desarrolladores integren experiencias que los niños utilicen. En Windows, eso empieza por la cuenta. La cuenta de Microsoft es una base de confianza a través de la cual Windows aporta protecciones y decisiones parentales adecuadas a la edad a más experiencias digitales que viven niños y adolescentes cada día.

La cuenta de un niño debería ayudar a desbloquear experiencias más seguras y adecuadas para la edad dondequiera que use Windows.

Esta es la primera de una serie de publicaciones sobre la seguridad familiar en Windows. En los próximos meses compartiré lo que construimos — en la cuenta, en la plataforma y en las herramientas que ofrecemos a padres, educadores y desarrolladores. Como Pavan compartió de manera reciente, la familia es un área clave en nuestro compromiso con mejorar la calidad de Windows. Construiremos a la luz de todos: lanzaremos nuevas capacidades, escucharemos con atención lo que nos dicen las familias y perfeccionaremos a medida que aprendemos.

De la cuenta a experiencias más seguras

Toda experiencia en Windows comienza con la identidad, y para los consumidores, esa base es la cuenta de Microsoft (MSA, por sus siglas en inglés).

MSA permite a Windows comprender el contexto clave sobre el usuario de manera coherente y respetuosa con la privacidad:

  • Ya sea niño, adolescente o adulto
  • Ya sea que se apliquen controles parentales o consentimiento
  • Si se ha completado la verificación de edad

Esto crea un modelo sencillo y escalable:

Cuenta de usuario → señal de edad → experiencia y controles apropiados para la edad

Dado que esta base está ligada a la cuenta, las protecciones están diseñadas para acompañar al usuario a través de Windows y aplicaciones, con la creación de experiencias más seguras de manera consistente, no solo en características individuales. Esto permite que las experiencias de producto apoyen la etapa de vida del usuario. Las protecciones se adaptan a niños, adolescentes y adultos.

Ampliar la seguridad a todo el ecosistema con las APIs de Windows Age

A medida que las experiencias se expanden a través de aplicaciones, servicios e IA, queda claro un reto: ¿cómo pueden estas experiencias comprender y respetar de manera consistente la edad de un usuario?

Para abordar esto, Windows introduce una nueva capacidad de plataforma: la API Windows Age.

Esta API aporta la conciencia sobre la edad más allá del sistema operativo al hacerla disponible en todo el ecosistema de Windows, de modo que las aplicaciones y servicios puedan ofrecer experiencias adecuadas a la edad utilizando la misma base de confianza. Al poner a disposición la concienciación sobre la edad como una capacidad de plataforma, Windows ayuda a los desarrolladores a crear salvaguardas en las experiencias desde el principio, en lugar de cargar la carga sobre niños y familias para gestionar las protecciones aplicación por aplicación.

La API de Windows Age ofrece las siguientes capacidades:

  • GetUserAgeRangeAsync: Proporciona categorías de grupos de edad no identificables de manera personal, para que las aplicaciones las utilicen para ofrecer experiencias de usuario seguras y adecuadas a su edad, al proteger los datos de fecha de nacimiento y la privacidad general de los usuarios. Los grupos de edad disponibles a través de la API incluyen: menores de 10, 10-12, 13-15, 16-17, 18+.
  • GetAgeVerificationStatusAsync: Proporciona estado de edad verificado para escenarios en los que la edad auto-reportada no es suficiente.
  • CheckAgeStatusAsync: Extiende la funcionalidad de la antigua API CheckUserAgeConsentGroupAsync para ofrecer clasificación de niños, menores o adultos a nivel global y según lo definido por las políticas regionales.

Estas señales están diseñadas para proteger la privacidad:

  • Las aplicaciones solo reciben la señal relacionada con la edad necesaria para la experiencia y no datos personales sensibles como la fecha completa de nacimiento
  • Los usuarios mantienen el control mediante el consentimiento y controles a nivel de plataforma
  • La información de edad se gestiona de manera coherente a nivel de plataforma para reducir el intercambio innecesario y la fragmentación

Esto permite un nuevo nivel de coherencia entre las experiencias de Windows. Las aplicaciones pueden reconocer cuándo un usuario es menor de cierta edad y ajustar las características en consecuencia, lo que ayuda a crear interacciones más seguras y adecuadas sin necesidad de configuración manual. Este enfoque de plataforma también ayuda a que las protecciones sean más consistentes y accesibles, de modo que las experiencias más seguras no dependan de si una familia descubre o configura controles en cada aplicación individual.

Por igual importante, esto se extiende a experiencias emergentes, incluida la IA, donde aplicar salvaguardas adecuadas a la edad es cada vez más importante a medida que la innovación sigue evolucionando.

Al convertir esto en una capacidad de plataforma, Windows fomenta una seguridad integrada y disponible en aplicaciones, servicios y nuevos escenarios.

Para más detalles, incluida la orientación para desarrolladores, consulten la documentación de la API de Windows Age que se encuentra en https://aka.ms/windows-age-api.

Construir hacia una garantía de edad con alta confianza

En muchos casos, entender la edad no es suficiente. Cada vez más, ciertas experiencias requieren una alta confianza en que el usuario es adulto.

Para apoyar esto, Microsoft amplía la garantía de edad a través de la cuenta Microsoft utilizando Microsoft Age Verification (MAV), una plataforma centralizada que permite a los usuarios verificar su edad una vez y usar ese estado en las experiencias de Microsoft. Los desarrolladores pueden acceder a este estado verificado a través de las APIs de Windows Age.

La verificación de edad ya está disponible en Microsoft Storefronts en Singapur, Brasil y Australia. A medida que más regiones del mundo amplían los requisitos regulatorios para la verificación de edad, Microsoft añadirá soporte para esas geografías y escenarios de productos.

MAV crea un modelo simple y consistente:

Verificar una vez → usarlo en todas partes

El estado verificado se almacena con la cuenta de Microsoft y puede usarse en Windows y aplicaciones para ofrecer experiencias adecuadas, lo que ayuda a reducir fricciones y mejorar la seguridad.

Dejar la seguridad más clara desde el principio

También mejoramos cómo se manifiestan los controles parentales durante la configuración del dispositivo, para que las familias puedan entender más fácil las protecciones disponibles y tomar decisiones informadas desde el principio.

En regiones como Francia y otras, Windows aumenta la visibilidad de los controles parentales y las experiencias de consentimiento durante la configuración, diseñada para aumentar la conciencia sobre las ofertas disponibles y permitir la aplicación temprana por parte de los padres.

Imagen de Family Safety integrado en Windows

Seguridad familiar y controles parentales

Con la Seguridad Familiar integrada en Windows, los padres pueden configurar el dispositivo de un niño y gestionar el tiempo frente a pantallas, las aplicaciones y el contenido en un solo lugar. Durante los últimos seis meses, nos hemos centrado en hacer que esos momentos cotidianos sean más fiables: aprobaciones parentales más rápidas, informes de actividades más claros y mayor control general, guiados por lo que los padres nos dijeron que era lo más importante. Pueden leer más sobre lo que ha mejorado y lo que viene a continuación aquí: Escuchar a las familias. Mejorar Microsoft Family.

Elevar el listón de la seguridad digital

La seguridad digital debe estar integrada en la plataforma, no añadirse más adelante.

Con la cuenta de Microsoft como base, e inversiones como la ampliación de la garantía de edad de la API Windows Age y controles parentales disponibles, Windows trabaja para apoyar un entorno digital más seguro que también sea adecuado para la edad, respetuoso con la privacidad y más fácil de navegar para las familias.

Nuestro objetivo es ayudar a crear experiencias en las que la seguridad, la privacidad y la participación puedan trabajar juntas en todo el ecosistema Windows.

Disponibilidad

Las APIs de Windows Age están ahora disponibles de manera amplia para Windows Insiders y pronto estarán disponibles para todos los usuarios de Windows. La API CheckAgeStatusAsync estará disponible con una actualización futura.

The post Ayudar a familias y educadores a apoyar experiencias más seguras en Windows appeared first on Source LATAM.

 

​The post Ayudar a familias y educadores a apoyar experiencias más seguras en Windows appeared first on Source LATAM.  

Publicado el — Deja un comentario

Qwen3.6-35B-A3B-NVFP4 and Wan2.1-T2V-1.3B-Diffusers models now available on Amazon SageMaker JumpStart

NVIDIA’s Qwen3.6-35B-A3B-NVFP4 and Alibaba’s Wan2.1-T2V-1.3B-Diffusers models are now available on Amazon SageMaker JumpStart, expanding the portfolio of foundation models available to AWS customers. These two models bring specialized capabilities spanning agentic coding with long-context reasoning and lightweight text-to-video generation, enabling customers to deploy high-performance, scalable AI solutions on AWS infrastructure.

These models address different enterprise AI challenges with specialized capabilities:

Qwen3.6-35B-A3B-NVFP4 is optimized for agentic coding, multimodal reasoning, and long-context understanding as the NVIDIA-quantized variant of Alibaba’s Qwen3.6-35B-A3B. This Mixture-of-Experts model contains 35B total parameters with only 3B activated per token (8 of 256 experts), supporting a 262K-token context window extendable to ~1M via YaRN scaling. Quantized to NVFP4 using NVIDIA’s ModelOpt framework, it preserves thinking across conversation turns, multi-token prediction, and tool calling for multi-step agent pipelines—all at a significantly reduced memory footprint.

Wan2.1-T2V-1.3B-Diffusers excels in text-to-video generation on consumer-grade hardware. Built on the diffusion transformer paradigm with a novel Video Variational Autoencoder (VAE), this 1.3B-parameter model generates high-quality, physics-consistent video clips from text prompts while requiring only 8.19 GB of VRAM. It can produce a 5-second 480p video on an RTX 4090 in approximately 4 minutes, making it one of the most accessible open-source video generation models available.

With SageMaker JumpStart, customers can deploy any of these models with just a few clicks to address their specific AI use cases.

To get started with these models, navigate to the SageMaker JumpStart model catalog in the SageMaker console or use the SageMaker Python SDK to deploy the models to your AWS account. For more information about deploying and using foundation models in SageMaker JumpStart, see the Amazon SageMaker JumpStart documentation.

 

​NVIDIA’s Qwen3.6-35B-A3B-NVFP4 and Alibaba’s Wan2.1-T2V-1.3B-Diffusers models are now available on Amazon SageMaker JumpStart, expanding the portfolio of foundation models available to AWS customers. These two models bring specialized capabilities spanning agentic coding with long-context reasoning and lightweight text-to-video generation, enabling customers to deploy high-performance, scalable AI solutions on AWS infrastructure.
These models address different enterprise AI challenges with specialized capabilities:
Qwen3.6-35B-A3B-NVFP4 is optimized for agentic coding, multimodal reasoning, and long-context understanding as the NVIDIA-quantized variant of Alibaba’s Qwen3.6-35B-A3B. This Mixture-of-Experts model contains 35B total parameters with only 3B activated per token (8 of 256 experts), supporting a 262K-token context window extendable to ~1M via YaRN scaling. Quantized to NVFP4 using NVIDIA’s ModelOpt framework, it preserves thinking across conversation turns, multi-token prediction, and tool calling for multi-step agent pipelines—all at a significantly reduced memory footprint.
Wan2.1-T2V-1.3B-Diffusers excels in text-to-video generation on consumer-grade hardware. Built on the diffusion transformer paradigm with a novel Video Variational Autoencoder (VAE), this 1.3B-parameter model generates high-quality, physics-consistent video clips from text prompts while requiring only 8.19 GB of VRAM. It can produce a 5-second 480p video on an RTX 4090 in approximately 4 minutes, making it one of the most accessible open-source video generation models available.
With SageMaker JumpStart, customers can deploy any of these models with just a few clicks to address their specific AI use cases.
To get started with these models, navigate to the SageMaker JumpStart model catalog in the SageMaker console or use the SageMaker Python SDK to deploy the models to your AWS account. For more information about deploying and using foundation models in SageMaker JumpStart, see the Amazon SageMaker JumpStart documentation.  

Publicado el — Deja un comentario

Ministral-3-3B-Instruct-2512 and Ministral-3-8B-Instruct-2512 models now available on Amazon SageMaker JumpStart

Mistral AI’s Ministral-3-3B-Instruct-2512 and Ministral-3-8B-Instruct-2512 models are now available on Amazon SageMaker JumpStart, expanding the portfolio of foundation models available to AWS customers. These two models from the Ministral 3 family bring compact, vision-capable language models purpose-built for edge deployment and resource-constrained environments, enabling customers to deploy high-performance, scalable AI solutions on AWS infrastructure.

These models address different enterprise AI challenges with specialized capabilities:

Ministral-3-3B-Instruct-2512 is engineered for ultra-lightweight edge deployment with multimodal understanding. Comprising a 3.4B language model and a 0.4B vision encoder, it fits in just 8GB of VRAM in FP8 while supporting a 256K-token context window. It offers vision analysis, multilingual instruction following across dozens of languages (including English, French, Spanish, German, Chinese, Japanese, Korean, and Arabic), strong system-prompt adherence, and native function calling with structured JSON output—all under the Apache 2.0 license.

Ministral-3-8B-Instruct-2512 delivers frontier-class capabilities comparable to its larger Mistral Small 3.2 24B counterpart in a compact 8B form factor. Built with an 8.4B language model and a 0.4B vision encoder, it fits in 12GB of VRAM in FP8 and features an interleaved sliding-window attention pattern for faster, memory-efficient inference. It shares the same vision, multilingual, agentic, and function-calling capabilities as its 3B sibling while offering stronger reasoning and generation performance.

With SageMaker JumpStart, customers can deploy any of these models with just a few clicks to address their specific AI use cases.

To get started with these models, navigate to the SageMaker JumpStart model catalog in the SageMaker console or use the SageMaker Python SDK to deploy the models to your AWS account. For more information about deploying and using foundation models in SageMaker JumpStart, see the Amazon SageMaker JumpStart documentation.

 

​Mistral AI’s Ministral-3-3B-Instruct-2512 and Ministral-3-8B-Instruct-2512 models are now available on Amazon SageMaker JumpStart, expanding the portfolio of foundation models available to AWS customers. These two models from the Ministral 3 family bring compact, vision-capable language models purpose-built for edge deployment and resource-constrained environments, enabling customers to deploy high-performance, scalable AI solutions on AWS infrastructure.
These models address different enterprise AI challenges with specialized capabilities:
Ministral-3-3B-Instruct-2512 is engineered for ultra-lightweight edge deployment with multimodal understanding. Comprising a 3.4B language model and a 0.4B vision encoder, it fits in just 8GB of VRAM in FP8 while supporting a 256K-token context window. It offers vision analysis, multilingual instruction following across dozens of languages (including English, French, Spanish, German, Chinese, Japanese, Korean, and Arabic), strong system-prompt adherence, and native function calling with structured JSON output—all under the Apache 2.0 license.
Ministral-3-8B-Instruct-2512 delivers frontier-class capabilities comparable to its larger Mistral Small 3.2 24B counterpart in a compact 8B form factor. Built with an 8.4B language model and a 0.4B vision encoder, it fits in 12GB of VRAM in FP8 and features an interleaved sliding-window attention pattern for faster, memory-efficient inference. It shares the same vision, multilingual, agentic, and function-calling capabilities as its 3B sibling while offering stronger reasoning and generation performance.
With SageMaker JumpStart, customers can deploy any of these models with just a few clicks to address their specific AI use cases.
To get started with these models, navigate to the SageMaker JumpStart model catalog in the SageMaker console or use the SageMaker Python SDK to deploy the models to your AWS account. For more information about deploying and using foundation models in SageMaker JumpStart, see the Amazon SageMaker JumpStart documentation.  

Publicado el — Deja un comentario

Gemma-4-31B-it-assistant and Gemma-4-31B-IT-NVFP4 models now available on Amazon SageMaker JumpStart

Google DeepMind’s Gemma-4-31B-it-assistant and NVIDIA’s Gemma-4-31B-IT-NVFP4 models are now available on Amazon SageMaker JumpStart, expanding the portfolio of foundation models available to AWS customers. These two models bring the flagship Gemma 4 31B dense architecture to enterprise workloads in both full-precision and optimized quantized variants, enabling customers to deploy high-performance, scalable AI solutions on AWS infrastructure.

These models address different enterprise AI challenges with specialized capabilities:

Gemma-4-31B-it-assistant is built for multimodal reasoning, coding, and agentic workflows as the assistant-tuned variant of Google’s flagship 31B dense model. It handles text and image inputs (including video as frame sequences) and generates text output, with a 256K-token context window and support for over 140 languages. Ranked #3 among open models on the Arena AI text leaderboard—outcompeting models 20x its size—it features a hybrid attention mechanism interleaving local sliding-window and full global attention with native function calling for building autonomous agents.

Gemma-4-31B-IT-NVFP4 delivers the same Gemma 4 31B capabilities at a fraction of the memory footprint. Quantized with NVIDIA’s ModelOpt framework to 4-bit FP4 precision, it reduces memory usage to ~18.5 GB (68% smaller than the base model) and achieves approximately 2.5x faster inference while retaining 97–99% of the original model’s quality. Ideal for cost-efficient, high-throughput production deployments on NVIDIA RTX, DGX Spark, and data center GPUs.

With SageMaker JumpStart, customers can deploy any of these models with just a few clicks to address their specific AI use cases.

To get started with these models, navigate to the SageMaker JumpStart model catalog in the SageMaker console or use the SageMaker Python SDK to deploy the models to your AWS account. For more information about deploying and using foundation models in SageMaker JumpStart, see the Amazon SageMaker JumpStart documentation.

 

​Google DeepMind’s Gemma-4-31B-it-assistant and NVIDIA’s Gemma-4-31B-IT-NVFP4 models are now available on Amazon SageMaker JumpStart, expanding the portfolio of foundation models available to AWS customers. These two models bring the flagship Gemma 4 31B dense architecture to enterprise workloads in both full-precision and optimized quantized variants, enabling customers to deploy high-performance, scalable AI solutions on AWS infrastructure.
These models address different enterprise AI challenges with specialized capabilities:
Gemma-4-31B-it-assistant is built for multimodal reasoning, coding, and agentic workflows as the assistant-tuned variant of Google’s flagship 31B dense model. It handles text and image inputs (including video as frame sequences) and generates text output, with a 256K-token context window and support for over 140 languages. Ranked #3 among open models on the Arena AI text leaderboard—outcompeting models 20x its size—it features a hybrid attention mechanism interleaving local sliding-window and full global attention with native function calling for building autonomous agents.
Gemma-4-31B-IT-NVFP4 delivers the same Gemma 4 31B capabilities at a fraction of the memory footprint. Quantized with NVIDIA’s ModelOpt framework to 4-bit FP4 precision, it reduces memory usage to ~18.5 GB (68% smaller than the base model) and achieves approximately 2.5x faster inference while retaining 97–99% of the original model’s quality. Ideal for cost-efficient, high-throughput production deployments on NVIDIA RTX, DGX Spark, and data center GPUs.
With SageMaker JumpStart, customers can deploy any of these models with just a few clicks to address their specific AI use cases.
To get started with these models, navigate to the SageMaker JumpStart model catalog in the SageMaker console or use the SageMaker Python SDK to deploy the models to your AWS account. For more information about deploying and using foundation models in SageMaker JumpStart, see the Amazon SageMaker JumpStart documentation.  

Publicado el — Deja un comentario

granite-speech-4.1-2b, kanana-2-30b-a3b-instruct, and OpenFold3 models now available on Amazon SageMaker JumpStart

BM’s granite-speech-4.1-2b, Kakao’s kanana-2-30b-a3b-instruct, and the OpenFold Consortium’s OpenFold3 models are now available on Amazon SageMaker JumpStart, expanding the portfolio of foundation models available to AWS customers. These three models bring specialized capabilities spanning multilingual speech recognition, bilingual agentic AI, and biomolecular structure prediction, enabling customers to deploy high-performance, scalable AI solutions on AWS infrastructure.

These models address different enterprise AI challenges with specialized capabilities:

granite-speech-4.1-2b is purpose-built for multilingual automatic speech recognition (ASR) and bidirectional speech translation (AST) across English, French, German, Spanish, Portuguese, and Japanese. This compact 2B-parameter speech-language model delivers a word error rate of 5.33% with a real-time factor of ~231, making it one of the most efficient ASR models in its class. Released under Apache 2.0, it integrates seamlessly into enterprise voice workflows for transcription, translation, and audio processing at scale.

kanana-2-30b-a3b-instruct excels in bilingual Korean-English instruction following and agentic AI workflows. Developed by Kakao, it adopts a cutting-edge architecture featuring Multi-head Latent Attention (MLA) and Mixture-of-Experts (MoE), activating only 3B of its 30B total parameters per forward pass for superior throughput. Post-trained with supervised fine-tuning and reinforcement learning, it supports up to 128K tokens via YaRN scaling and is designed to function as an AI collaborator that understands context and acts proactively.

OpenFold3 provides all-atom biomolecular complex structure prediction for proteins, DNA, RNA, and small-molecule ligands. Developed by the OpenFold Consortium and the AlQuraishi Lab at Columbia University, this diffusion-based model extends structure prediction beyond single proteins to model multi-chain complexes and heterogeneous biomolecular interactions. It supports computer-aided drug design and is applicable across academic and pharmaceutical research labs.

With SageMaker JumpStart, customers can deploy any of these models with just a few clicks to address their specific AI use cases.

To get started with these models, navigate to the SageMaker JumpStart model catalog in the SageMaker console or use the SageMaker Python SDK to deploy the models to your AWS account. For more information about deploying and using foundation models in SageMaker JumpStart, see the Amazon SageMaker JumpStart documentation.

 

​BM’s granite-speech-4.1-2b, Kakao’s kanana-2-30b-a3b-instruct, and the OpenFold Consortium’s OpenFold3 models are now available on Amazon SageMaker JumpStart, expanding the portfolio of foundation models available to AWS customers. These three models bring specialized capabilities spanning multilingual speech recognition, bilingual agentic AI, and biomolecular structure prediction, enabling customers to deploy high-performance, scalable AI solutions on AWS infrastructure.
These models address different enterprise AI challenges with specialized capabilities:
granite-speech-4.1-2b is purpose-built for multilingual automatic speech recognition (ASR) and bidirectional speech translation (AST) across English, French, German, Spanish, Portuguese, and Japanese. This compact 2B-parameter speech-language model delivers a word error rate of 5.33% with a real-time factor of ~231, making it one of the most efficient ASR models in its class. Released under Apache 2.0, it integrates seamlessly into enterprise voice workflows for transcription, translation, and audio processing at scale.
kanana-2-30b-a3b-instruct excels in bilingual Korean-English instruction following and agentic AI workflows. Developed by Kakao, it adopts a cutting-edge architecture featuring Multi-head Latent Attention (MLA) and Mixture-of-Experts (MoE), activating only 3B of its 30B total parameters per forward pass for superior throughput. Post-trained with supervised fine-tuning and reinforcement learning, it supports up to 128K tokens via YaRN scaling and is designed to function as an AI collaborator that understands context and acts proactively.
OpenFold3 provides all-atom biomolecular complex structure prediction for proteins, DNA, RNA, and small-molecule ligands. Developed by the OpenFold Consortium and the AlQuraishi Lab at Columbia University, this diffusion-based model extends structure prediction beyond single proteins to model multi-chain complexes and heterogeneous biomolecular interactions. It supports computer-aided drug design and is applicable across academic and pharmaceutical research labs.
With SageMaker JumpStart, customers can deploy any of these models with just a few clicks to address their specific AI use cases.
To get started with these models, navigate to the SageMaker JumpStart model catalog in the SageMaker console or use the SageMaker Python SDK to deploy the models to your AWS account. For more information about deploying and using foundation models in SageMaker JumpStart, see the Amazon SageMaker JumpStart documentation.  

Publicado el — Deja un comentario

Presentamos Azure Multicloud Interconnect para AWS

Presentamos Azure Multicloud Interconnect para AWS

Gráfico de Azure Multicloud Interconnect

Por: Narayan Annamalai, vicepresidente y jefe de productos de Azure Networking Services.

A medida que las organizaciones aceleran la adopción de la IA y modernizan sus entornos digitales, las aplicaciones, datos e infraestructuras abarcan cada vez más múltiples entornos en la nube. Los clientes eligen la mejor plataforma para cada carga de trabajo, para aprovechar capacidades únicas de los proveedores y así impulsar la innovación, la resiliencia y la agilidad empresarial.

Sin embargo, aunque las estrategias de nube múltiple se han vuelto habituales, la red entre entornos de nube es todavía compleja. Establecer conectividad privada suele requerir que los clientes unan de manera manual los servicios, coordinen el aprovisionamiento entre proveedores, gestionen múltiples procesos operativos y naveguen por experiencias de soporte fragmentadas. Lo que debería ser una decisión sencilla de conectividad puede llevar semanas o meses en implementarse y gestionarse.

Exploren Azure Multicloud Interconnect

Hoy en día, Microsoft y Amazon Web Services (AWS) dan un paso importante para simplificar esa experiencia.

Microsoft Azure y AWS están entusiasmados por colaborar en una solución de red de nube múltiple que utilice tanto AWS Interconnect – multicloud como Azure Multicloud Interconnect para la interoperabilidad de red, lo que permite a los clientes establecer una conectividad privada y de alto rendimiento entre Microsoft Azure y AWS mediante una experiencia simplificada y nativa de la nube.

Esta colaboración se construye a través de las especificaciones Open API para la interoperabilidad de la red. Azure Multicloud Interconnect ayuda a eliminar gran parte de la complejidad asociada de manera tradicional a las redes de nube múltiple. Los clientes pueden proporcionar conectividad a través de una experiencia integrada mientras se benefician de rendimiento, resiliencia, seguridad y simplicidad operativa de nivel empresarial.

Simplificación de la conectividad multicloud

Hasta ahora, las organizaciones que conectaban entornos Azure y AWS requerían una planificación cuidadosa, conectividad física, configuración de enrutamiento, coordinación de provisionamiento, monitorización y gestión del ciclo de vida para ensamblar y mantener múltiples componentes entre proveedores.

Azure Multicloud Interconnect cambia de manera fundamental este modelo. Con la colaboración de Microsoft y AWS para utilizar la especificación estandarizada de Open API, los clientes pueden establecer conectividad privada dedicada mediante una experiencia simplificada que abstrae la complejidad subyacente de las redes multicloud. En lugar de centrarse en la gestión de infraestructuras, las organizaciones pueden centrarse en entregar aplicaciones, mover datos y acelerar los resultados empresariales.

El resultado es un camino de alto ancho de banda y más predecible para desplegar arquitecturas de nube múltiple para cargas de trabajo críticas para la misión.

Diseñado para la era de la IA

Las cargas de trabajo de entrenamiento e inferencia requieren con frecuencia de acceso a datos distribuidos entre entornos. Las empresas diseñan cada vez más aplicaciones que cruzan los límites de la nube para mantener los requisitos de rendimiento, seguridad y cumplimiento.

Azure Multicloud Interconnect está diseñado para soportar estos requisitos en evolución con una conectividad privada de alta capacidad que se extiende hasta Azure Private Link, para brindar un camino privado de extremo a extremo entre las nubes.

Esta combinación de conectividad de alto rendimiento y simplicidad operativa permite a los clientes avanzar más rápido mientras desarrollan la próxima generación de aplicaciones y servicios habilitados por IA.

A medida que la IA transforma todos los sectores, los clientes necesitan la libertad de colocar datos, aplicaciones e infraestructura donde sea que aporte el mayor valor empresarial. Azure Multicloud Interconnect ayuda a hacerlo posible al proporcionar una conectividad privada resiliente y de alto rendimiento entre Azure y AWS mediante una experiencia simplificada y nativa en la nube. Juntos, redujimos la complejidad de las redes de nube múltiple y proporcionamos a los clientes la escala, fiabilidad y agilidad que necesitan para impulsar la próxima generación de innovación basada en IA y datos.

—Robert Kennedy, vicepresidente de Servicios de Red en AWS

Avanzar hacia un futuro de nube múltiple abierta

Azure Multicloud Interconnect representa más que una nueva oferta de conectividad: es un paso hacia un ecosistema de nube más abierto e interconectado.

De cara al futuro, vemos la oportunidad de ampliar este modelo más allá de una relación nube a nube. La misma especificación abierta de API puede ayudar a permitir una interoperabilidad más amplia entre proveedores de nube hiperescaladas, para crear una experiencia más coherente para los clientes que operan en entornos de nube múltiple cada vez más diversos.

Los clientes pueden desplegar conectividad a velocidades de hasta 100 Gbps desde  el primer día en disponibilidad general, lo que ayuda a satisfacer las necesidades de aplicaciones de alto ancho de banda y proporciona la base para el crecimiento futuro. A medida que aumenta la demanda, la capacidad puede expandirse de manera dinámica, lo que permite a las organizaciones escalar sin interrumpir las operaciones ni rediseñar la arquitectura de su red.

Más allá de los hiperescaladores, este enfoque tiene el potencial de simplificar la conectividad con los proveedores de servicios de red y los operadores de telecomunicaciones. Al adoptar un modelo común de interoperabilidad, los proveedores de la nube y los operadores pueden trabajar juntos para agilizar la conectividad de última milla, acelerar el aprovisionamiento y reducir la complejidad operativa a lo largo del recorrido integral del cliente.

Nuestra visión a largo plazo es un ecosistema abierto donde los hiperescaladores, proveedores de servicios de red, y operadores de telecomunicaciones utilicen un marco común de interoperabilidad para establecer y operar la conectividad mediante APIs estandarizadas. Los clientes deberían poder proporcionar conectividad confiable y de alto rendimiento entre nubes, redes metropolitanas y ubicaciones empresariales con la misma simplicidad y automatización que esperan de los servicios modernos en la nube.

Para saber más sobre cómo implementar redes multicloud en Azure, por favor visiten el blog detallado o la página de Microsoft Learn para empezar. Para leer el anuncio de AWS, visiten su blog.

The post Presentamos Azure Multicloud Interconnect para AWS appeared first on Source LATAM.

 

​The post Presentamos Azure Multicloud Interconnect para AWS appeared first on Source LATAM.  

Publicado el — Deja un comentario

AWS Elemental MediaLive enables frame-accurate pipeline locking for streams without timecode

AWS Elemental MediaLive now supports Video Aligned Locking, a new feature to synchronize video pipelines without requiring timecode from the source. Previously, achieving frame-accurate locking across video outputs required investing in specialized hardware or managing complex, external synchronization workflows. While timecode is generally standard for highly produced, traditional broadcast content, it is often unavailable and difficult to manage in typical digital streaming workflows. The new capability uses visual signatures to automatically identify and align specific frames across multiple video streams, enabling customers to achieve frame-accurate input switching across standard pipeline channels, as well as linked, cross-region single-pipeline channels.

Video Aligned Locking supports outputs including HLS, MediaPackage, CMAF Ingest, UDP and SRT. To learn more about how to configure this feature, visit the Requirements for Video Aligned Locking documentation. For a deeper dive into channel synchronization, visit the guide on Implementing Pipeline Locking.

 

​AWS Elemental MediaLive now supports Video Aligned Locking, a new feature to synchronize video pipelines without requiring timecode from the source. Previously, achieving frame-accurate locking across video outputs required investing in specialized hardware or managing complex, external synchronization workflows. While timecode is generally standard for highly produced, traditional broadcast content, it is often unavailable and difficult to manage in typical digital streaming workflows. The new capability uses visual signatures to automatically identify and align specific frames across multiple video streams, enabling customers to achieve frame-accurate input switching across standard pipeline channels, as well as linked, cross-region single-pipeline channels.
Video Aligned Locking supports outputs including HLS, MediaPackage, CMAF Ingest, UDP and SRT. To learn more about how to configure this feature, visit the Requirements for Video Aligned Locking documentation. For a deeper dive into channel synchronization, visit the guide on Implementing Pipeline Locking.  

Publicado el — Deja un comentario

AWS Lambda now supports direct read configuration for Amazon S3 Files

AWS Lambda now supports direct read configuration for Amazon S3 Files, letting you configure which storage your functions read from: S3 Files high-performance storage or your S3 bucket. With this launch, you can optimize the throughput and latency of file reads for your Lambda functions based on your application requirements.

Customers use S3 Files with Lambda functions to build scalable data processing pipelines and stateful agentic workloads with the performance and simplicity of a file system, while benefiting from the scalability, durability, and cost-effectiveness of S3. S3 Files serves data from high-performance storage for low latency or directly from your S3 bucket for high throughput on large reads, automatically routing each operation to the storage best suited for it. By default, Lambda supports direct reads from your S3 bucket only for functions configured with 512 MB of memory or higher. However, without control over the direct read configuration, you cannot optimize read performance for your specific application requirements. With this launch, you can explicitly enable or disable direct read for S3 Files on your Lambda functions, independent of function memory size. When you enable direct read, your function reads files 1 MB or larger directly from your S3 bucket for maximum throughput, and smaller files are served through the high-performance storage. When you disable it, all reads are served through the high-performance storage for the lowest latency.

This capability is available in all AWS commercial Regions, AWS GovCloud (US-East), and AWS GovCloud (US-West) Regions, except Asia Pacific (New Zealand), Middle East (Bahrain), and Middle East (UAE). You can configure direct read for S3 Files using the AWS Management Console, AWS CLI, AWS SDKs, and AWS CloudFormation. There is no additional charge beyond standard Lambda and S3 Files pricing. To learn more about how to use S3 Files with your Lambda function, visit the AWS Lambda developer guide.

 

​AWS Lambda now supports direct read configuration for Amazon S3 Files, letting you configure which storage your functions read from: S3 Files high-performance storage or your S3 bucket. With this launch, you can optimize the throughput and latency of file reads for your Lambda functions based on your application requirements.
Customers use S3 Files with Lambda functions to build scalable data processing pipelines and stateful agentic workloads with the performance and simplicity of a file system, while benefiting from the scalability, durability, and cost-effectiveness of S3. S3 Files serves data from high-performance storage for low latency or directly from your S3 bucket for high throughput on large reads, automatically routing each operation to the storage best suited for it. By default, Lambda supports direct reads from your S3 bucket only for functions configured with 512 MB of memory or higher. However, without control over the direct read configuration, you cannot optimize read performance for your specific application requirements. With this launch, you can explicitly enable or disable direct read for S3 Files on your Lambda functions, independent of function memory size. When you enable direct read, your function reads files 1 MB or larger directly from your S3 bucket for maximum throughput, and smaller files are served through the high-performance storage. When you disable it, all reads are served through the high-performance storage for the lowest latency.
This capability is available in all AWS commercial Regions, AWS GovCloud (US-East), and AWS GovCloud (US-West) Regions, except Asia Pacific (New Zealand), Middle East (Bahrain), and Middle East (UAE). You can configure direct read for S3 Files using the AWS Management Console, AWS CLI, AWS SDKs, and AWS CloudFormation. There is no additional charge beyond standard Lambda and S3 Files pricing. To learn more about how to use S3 Files with your Lambda function, visit the AWS Lambda developer guide.  

Publicado el — Deja un comentario

Amazon EC2 X2idn instances are now available in Asia Pacific (Hong Kong)

Memory-optimized Amazon Elastic Compute Cloud (Amazon EC2) X2idn instances are now available in Asia Pacific (Hong Kong) Region. These instances, powered by 3rd generation Intel Xeon Scalable Processors and built with AWS Nitro System, are designed for memory-intensive workloads and deliver improvements in performance compared to previous generation X1 instances. These instances are SAP-certified for running Business Suite on HANA, SAP S/4HANA, Data Mart Solutions on HANA, Business Warehouse on HANA, SAP BW/4HANA, and SAP NetWeaver workloads on any database.

 

​Memory-optimized Amazon Elastic Compute Cloud (Amazon EC2) X2idn instances are now available in Asia Pacific (Hong Kong) Region. These instances, powered by 3rd generation Intel Xeon Scalable Processors and built with AWS Nitro System, are designed for memory-intensive workloads and deliver improvements in performance compared to previous generation X1 instances. These instances are SAP-certified for running Business Suite on HANA, SAP S/4HANA, Data Mart Solutions on HANA, Business Warehouse on HANA, SAP BW/4HANA, and SAP NetWeaver workloads on any database.  

Publicado el — Deja un comentario

Amazon SageMaker HyperPod now supports model caching for faster inference autoscaling and reduced cold starts

Amazon SageMaker HyperPod now supports model caching, an inference optimization that pre-loads model weights and container images onto cluster nodes so pods start in seconds instead of minutes.

When running LLM inference at scale for workloads like chat assistants, agentic pipelines, RAG, and document analysis, cold start is a real bottleneck. Deployments and scale-out events spend most of their time downloading container images and model weights. As model size increases, this gets worse, with large models taking tens of minutes before they can serve traffic.

Model caching solves this with two independent capabilities. The weights cache stores model weights on local NVMe so pods read from fast local storage instead of pulling from S3 or FSx over the network. The image cache pre-pulls the container image so pods skip the ECR download entirely. If a pod lands on a node without a warm cache, it falls back to pulling from the original source automatically, so there is no risk of pods getting stuck or failing.

Benchmarks across models from 57 GB to 145 GB show around 60% faster scale-out, and the image cache cuts over two minutes of image-pull time (97% reduction). The benefit grows with model size while retaining the reliability of the original source path.

Customers enable model caching through the HyperPod Inference Operator by adding a modelCacheConfig section to their InferenceEndpointConfig or JumpStartModel resource. The operator handles the full lifecycle with no manual setup or cleanup.

Model caching is now generally available in all regions where SageMaker HyperPod is available. To get started, see the SageMaker HyperPod documentation.

 

 

​Amazon SageMaker HyperPod now supports model caching, an inference optimization that pre-loads model weights and container images onto cluster nodes so pods start in seconds instead of minutes.
When running LLM inference at scale for workloads like chat assistants, agentic pipelines, RAG, and document analysis, cold start is a real bottleneck. Deployments and scale-out events spend most of their time downloading container images and model weights. As model size increases, this gets worse, with large models taking tens of minutes before they can serve traffic.
Model caching solves this with two independent capabilities. The weights cache stores model weights on local NVMe so pods read from fast local storage instead of pulling from S3 or FSx over the network. The image cache pre-pulls the container image so pods skip the ECR download entirely. If a pod lands on a node without a warm cache, it falls back to pulling from the original source automatically, so there is no risk of pods getting stuck or failing.
Benchmarks across models from 57 GB to 145 GB show around 60% faster scale-out, and the image cache cuts over two minutes of image-pull time (97% reduction). The benefit grows with model size while retaining the reliability of the original source path.
Customers enable model caching through the HyperPod Inference Operator by adding a modelCacheConfig section to their InferenceEndpointConfig or JumpStartModel resource. The operator handles the full lifecycle with no manual setup or cleanup.
Model caching is now generally available in all regions where SageMaker HyperPod is available. To get started, see the SageMaker HyperPod documentation.