Amazon SageMaker AI now supports Amazon EC2 G6e instances for batch transform. Batch Transform enables you to run predictions on datasets stored in Amazon S3 and is suited for large datasets that do not require a persistent inference endpoint.
Amazon EC2 G6e instances are powered by up to eight NVIDIA L40S Tensor Core GPUs with 48 GB of memory per GPU and third-generation AMD EPYC processors. G6e instances deliver improved performance for GPU-intensive workloads. With this launch, you can use G6e instances for GPU-intensive offline inference workloads, including large language models and diffusion models that generate images, video, and audio. To get started, select a supported ml.g6e instance type when creating a Batch Transform job through AWS SDKs, AWS CLI, or the CreateTransformJob API.
G6e support for Batch Transform is available in US East (N. Virginia), US East (Ohio), US West (Oregon), Asia Pacific (Mumbai), and Asia Pacific (Hyderabad). To learn more, visit the Amazon SageMaker AI product page and see the Batch Transform documentation. For pricing information on these instances, please visit our pricing page.
Amazon SageMaker AI now supports Amazon EC2 G6e instances for batch transform. Batch Transform enables you to run predictions on datasets stored in Amazon S3 and is suited for large datasets that do not require a persistent inference endpoint.
Amazon EC2 G6e instances are powered by up to eight NVIDIA L40S Tensor Core GPUs with 48 GB of memory per GPU and third-generation AMD EPYC processors. G6e instances deliver improved performance for GPU-intensive workloads. With this launch, you can use G6e instances for GPU-intensive offline inference workloads, including large language models and diffusion models that generate images, video, and audio. To get started, select a supported ml.g6e instance type when creating a Batch Transform job through AWS SDKs, AWS CLI, or the CreateTransformJob API.
G6e support for Batch Transform is available in US East (N. Virginia), US East (Ohio), US West (Oregon), Asia Pacific (Mumbai), and Asia Pacific (Hyderabad). To learn more, visit the Amazon SageMaker AI product page and see the Batch Transform documentation. For pricing information on these instances, please visit our pricing page.
Today, AWS Model Context Protocol Server (AWS MCP Server) added a serverless capability so that coding agents such as Claude Code and Kiro can efficiently diagnose issues with your Lambda functions. The serverless capability helps you troubleshoot your running Lambda functions and their connected resources.
The AWS MCP Server, available through the Agent Toolkit for AWS or as a standalone installation, is a managed service that gives AI coding agents secure access to AWS services. With the new AWS MCP Sever serverless capability, your coding agent inspects your Lambda function and its connected resources across Amazon API Gateway, Amazon EventBridge, Amazon S3, Amazon DynamoDB, Amazon SNS, Amazon SQS, and AWS Step Functions. The agent can correlate error signals against a 7-day baseline to pinpoint what changed, surface recurring errors to identify trends, retrieve the deployed configuration of your function and connected resources, provide a timeline of recent changes to track what happened, and analyze service latency across connected resources. As the agent gets comprehensive data in a single call, it consumes fewer tokens compared to orchestrating multiple API calls.
To get started, configure the Agent toolkit for AWS by running ‘aws configure agent-toolkit’ from the AWS CLI, or enable the AWS MCP Server directly.
The AWS MCP Server can access services in all commercial AWS Regions, while the AWS MCP Server itself runs in the US East (N. Virginia) and Europe (Frankfurt) Regions. The serverless diagnostic capabilities in the AWS MCP Server are available at no additional cost. To learn more, see the user guide.
Today, AWS Model Context Protocol Server (AWS MCP Server) added a serverless capability so that coding agents such as Claude Code and Kiro can efficiently diagnose issues with your Lambda functions. The serverless capability helps you troubleshoot your running Lambda functions and their connected resources.
The AWS MCP Server, available through the Agent Toolkit for AWS or as a standalone installation, is a managed service that gives AI coding agents secure access to AWS services. With the new AWS MCP Sever serverless capability, your coding agent inspects your Lambda function and its connected resources across Amazon API Gateway, Amazon EventBridge, Amazon S3, Amazon DynamoDB, Amazon SNS, Amazon SQS, and AWS Step Functions. The agent can correlate error signals against a 7-day baseline to pinpoint what changed, surface recurring errors to identify trends, retrieve the deployed configuration of your function and connected resources, provide a timeline of recent changes to track what happened, and analyze service latency across connected resources. As the agent gets comprehensive data in a single call, it consumes fewer tokens compared to orchestrating multiple API calls.
To get started, configure the Agent toolkit for AWS by running ‘aws configure agent-toolkit’ from the AWS CLI, or enable the AWS MCP Server directly.
The AWS MCP Server can access services in all commercial AWS Regions, while the AWS MCP Server itself runs in the US East (N. Virginia) and Europe (Frankfurt) Regions. The serverless diagnostic capabilities in the AWS MCP Server are available at no additional cost. To learn more, see the user guide.
Starting today, Amazon Elastic Compute Cloud (Amazon EC2) C8g instances are available in AWS Asia Pacific (Taipei, New Zealand), and AWS GovCloud (US-East) regions. These instances are powered by AWS Graviton4 processors and deliver up to 30% better performance compared to AWS Graviton3-based instances. Amazon EC2 C8g instances are built for compute-intensive workloads, such as high performance computing (HPC), batch processing, gaming, video encoding, scientific modeling, distributed analytics, CPU-based machine learning (ML) inference, and ad serving. These instances are built on the AWS Nitro System, which offloads CPU virtualization, storage, and networking functions to dedicated hardware and software to enhance the performance and security of your workloads.
AWS Graviton4-based Amazon EC2 instances deliver the best performance and energy efficiency for a broad range of workloads running on Amazon EC2. These instances offer larger instance sizes with up to 3x more vCPUs and memory compared to Graviton3-based Amazon C7g instances. AWS Graviton4 processors are up to 40% faster for databases, 30% faster for web applications, and 45% faster for large Java applications than AWS Graviton3 processors. C8g instances are available in 12 different instance sizes, including two bare metal sizes. They offer up to 50 Gbps enhanced networking bandwidth and up to 40 Gbps of bandwidth to the Amazon Elastic Block Store (Amazon EBS).
Starting today, Amazon Elastic Compute Cloud (Amazon EC2) C8g instances are available in AWS Asia Pacific (Taipei, New Zealand), and AWS GovCloud (US-East) regions. These instances are powered by AWS Graviton4 processors and deliver up to 30% better performance compared to AWS Graviton3-based instances. Amazon EC2 C8g instances are built for compute-intensive workloads, such as high performance computing (HPC), batch processing, gaming, video encoding, scientific modeling, distributed analytics, CPU-based machine learning (ML) inference, and ad serving. These instances are built on the AWS Nitro System, which offloads CPU virtualization, storage, and networking functions to dedicated hardware and software to enhance the performance and security of your workloads. AWS Graviton4-based Amazon EC2 instances deliver the best performance and energy efficiency for a broad range of workloads running on Amazon EC2. These instances offer larger instance sizes with up to 3x more vCPUs and memory compared to Graviton3-based Amazon C7g instances. AWS Graviton4 processors are up to 40% faster for databases, 30% faster for web applications, and 45% faster for large Java applications than AWS Graviton3 processors. C8g instances are available in 12 different instance sizes, including two bare metal sizes. They offer up to 50 Gbps enhanced networking bandwidth and up to 40 Gbps of bandwidth to the Amazon Elastic Block Store (Amazon EBS). To learn more, see Amazon EC2 C8g Instances. To explore how to migrate your workloads to Graviton-based instances, see AWS Graviton Fast Start program and Porting Advisor for Graviton. To get started, see the AWS Management Console.
Starting today, Amazon Elastic Compute Cloud (Amazon EC2) C9g and C9gd instances, powered by AWS Graviton5 processors, are available in the Asia Pacific (Tokyo) region. AWS Graviton5 processors are the fifth generation of custom-designed CPUs, delivering the best price performance for compute-intensive workloads running on Amazon EC2.
C9g and C9gd instances deliver up to 25% better compute performance compared to AWS Graviton4-based C8g and C8gd instances. They are up to 30% faster for databases, up to 35% faster for web applications, and up to 35% faster for machine learning. These instances are built on the sixth-generation AWS Nitro System and are the first to feature the Nitro Isolation Engine, harnessing formal verification to provide mathematical assurance that customer workloads are isolated from each other and AWS operators, pioneering a new standard for mathematically proven cloud security.
C9g instances are ideal for workloads such as high-performance computing (HPC), batch processing, gaming, video encoding, scientific modeling, distributed analytics, CPU-based machine learning (ML) inference, real time analytics, and ad serving. C9gd instances offer local NVMe-based SSD block-level storage for customers running compute-intensive workloads that also require high-speed, low-latency local storage for scratch space, temporary files, and caches.
Starting today, Amazon Elastic Compute Cloud (Amazon EC2) C9g and C9gd instances, powered by AWS Graviton5 processors, are available in the Asia Pacific (Tokyo) region. AWS Graviton5 processors are the fifth generation of custom-designed CPUs, delivering the best price performance for compute-intensive workloads running on Amazon EC2.
C9g and C9gd instances deliver up to 25% better compute performance compared to AWS Graviton4-based C8g and C8gd instances. They are up to 30% faster for databases, up to 35% faster for web applications, and up to 35% faster for machine learning. These instances are built on the sixth-generation AWS Nitro System and are the first to feature the Nitro Isolation Engine, harnessing formal verification to provide mathematical assurance that customer workloads are isolated from each other and AWS operators, pioneering a new standard for mathematically proven cloud security.
C9g instances are ideal for workloads such as high-performance computing (HPC), batch processing, gaming, video encoding, scientific modeling, distributed analytics, CPU-based machine learning (ML) inference, real time analytics, and ad serving. C9gd instances offer local NVMe-based SSD block-level storage for customers running compute-intensive workloads that also require high-speed, low-latency local storage for scratch space, temporary files, and caches.
C9g and C9gd instances are available for purchase in Asia Pacific (Tokyo) via Savings Plans, On-Demand, Spot instances, Dedicated instances, or Dedicated hosts. Level up your compute with AWS Graviton and get started today.
AWS Transfer Family SFTP Connectors now continue running file transfers while you rotate the credentials used to authenticate with remote SFTP servers. You no longer need to update the connector to point to a new secret version each time a credential rotates, removing a manual step and helping avoid failed transfers during the rotation window.
Connectors can now retrieve credentials from an ordered list of AWS Secrets Manager version stages, such as the current and previous versions, during authentication. The connector automatically tries each version in the order you specify and proceeds with the first that succeeds, so transfers continue uninterrupted as credentials are rotated on either side. You configure this when you create or update a connector, and the connector’s credentials must be stored in AWS Secrets Manager.
Support for continuing transfers during credential rotation is available in all AWS Regions where AWS Transfer Family SFTP Connectors are supported. To learn more, visit the AWS Transfer Family User Guide. Get started with AWS Transfer Family in the AWS Transfer Family console.
AWS Transfer Family SFTP Connectors now continue running file transfers while you rotate the credentials used to authenticate with remote SFTP servers. You no longer need to update the connector to point to a new secret version each time a credential rotates, removing a manual step and helping avoid failed transfers during the rotation window.
Connectors can now retrieve credentials from an ordered list of AWS Secrets Manager version stages, such as the current and previous versions, during authentication. The connector automatically tries each version in the order you specify and proceeds with the first that succeeds, so transfers continue uninterrupted as credentials are rotated on either side. You configure this when you create or update a connector, and the connector’s credentials must be stored in AWS Secrets Manager.
Support for continuing transfers during credential rotation is available in all AWS Regions where AWS Transfer Family SFTP Connectors are supported. To learn more, visit the AWS Transfer Family User Guide. Get started with AWS Transfer Family in the AWS Transfer Family console.
Amazon Elastic Container Service (Amazon ECS) now support non-critical Managed Daemons for ECS Managed Instances. You can configure a daemon as non-critical so that your mission-critical application tasks continue running uninterrupted, even when a daemon fails, stops, or becomes unhealthy.
Amazon ECS Managed Daemons lets you centrally deploy and manage software agents independently of application deployments to ensure consistent coverage and compliance across your container infrastructure. For some mission-critical applications, uninterrupted execution of application tasks may matter more than auxiliary daemon functionality like logging or metrics collection. With this launch, you can now configure a managed daemon as non-critical so that its failure does not cause your application tasks to churn. When a non-critical daemon task fails, stops, or becomes unhealthy, the container instance remains active, your existing application tasks continue running uninterrupted, ECS continues placing new application tasks on it, and instance registration is never blocked, so your tasks launch immediately even when the daemon fails to start. Amazon ECS emits an EventBridge event when a daemon task fails to start and records service action logs for both critical and non-critical daemons, giving you full observability into daemon health.
To get started, you can use AWS Console, CLI, CloudFormation, or AWS SDKs to create a non-critical daemon by setting the critical parameter to false when creating or updating a daemon. Non-critical daemons are available in all AWS Regions where Amazon ECS Managed Daemons are supported. To learn more, refer to our documentation page.
Amazon Elastic Container Service (Amazon ECS) now support non-critical Managed Daemons for ECS Managed Instances. You can configure a daemon as non-critical so that your mission-critical application tasks continue running uninterrupted, even when a daemon fails, stops, or becomes unhealthy.
Amazon ECS Managed Daemons lets you centrally deploy and manage software agents independently of application deployments to ensure consistent coverage and compliance across your container infrastructure. For some mission-critical applications, uninterrupted execution of application tasks may matter more than auxiliary daemon functionality like logging or metrics collection. With this launch, you can now configure a managed daemon as non-critical so that its failure does not cause your application tasks to churn. When a non-critical daemon task fails, stops, or becomes unhealthy, the container instance remains active, your existing application tasks continue running uninterrupted, ECS continues placing new application tasks on it, and instance registration is never blocked, so your tasks launch immediately even when the daemon fails to start. Amazon ECS emits an EventBridge event when a daemon task fails to start and records service action logs for both critical and non-critical daemons, giving you full observability into daemon health.
To get started, you can use AWS Console, CLI, CloudFormation, or AWS SDKs to create a non-critical daemon by setting the critical parameter to false when creating or updating a daemon. Non-critical daemons are available in all AWS Regions where Amazon ECS Managed Daemons are supported. To learn more, refer to our documentation page.
Starting today, Amazon Elastic Compute Cloud (Amazon EC2) P6-B200 instances accelerated by NVIDIA Blackwell GPUs are available in the AWS Asia Pacific (Hyderabad) Region. These instances offer up to 2x performance compared to P5en instances for AI training and inference.
P6-B200 instances feature 8 Blackwell GPUs with 1440 GB of high-bandwidth GPU memory and a 60% increase in GPU memory bandwidth compared to P5en, 5th Generation Intel Xeon processors (Emerald Rapids), and up to 3.2 terabits per second of Elastic Fabric Adapter (EFAv4) networking. P6-B200 instances are powered by the AWS Nitro System, so you can reliably and securely scale AI workloads within Amazon EC2 UltraClusters to tens of thousands of GPUs.
P6-B200 instances are now available in p6-b200.48xlarge size in the following AWS Regions: US West (Oregon), US East (N. Virginia, Ohio), AWS GovCloud (US-West, US-East), and Asia Pacific (Hyderabad, Mumbai) Regions. To learn more about P6-B200 instances, visit Amazon EC2 P6 instances.
Starting today, Amazon Elastic Compute Cloud (Amazon EC2) P6-B200 instances accelerated by NVIDIA Blackwell GPUs are available in the AWS Asia Pacific (Hyderabad) Region. These instances offer up to 2x performance compared to P5en instances for AI training and inference.
P6-B200 instances feature 8 Blackwell GPUs with 1440 GB of high-bandwidth GPU memory and a 60% increase in GPU memory bandwidth compared to P5en, 5th Generation Intel Xeon processors (Emerald Rapids), and up to 3.2 terabits per second of Elastic Fabric Adapter (EFAv4) networking. P6-B200 instances are powered by the AWS Nitro System, so you can reliably and securely scale AI workloads within Amazon EC2 UltraClusters to tens of thousands of GPUs.
P6-B200 instances are now available in p6-b200.48xlarge size in the following AWS Regions: US West (Oregon), US East (N. Virginia, Ohio), AWS GovCloud (US-West, US-East), and Asia Pacific (Hyderabad, Mumbai) Regions. To learn more about P6-B200 instances, visit Amazon EC2 P6 instances.
Starting today, Amazon Elastic Cloud Compute (Amazon EC2) P6-B300 instances are available in the AWS Asia Pacific (Jakarta) Region. P6-B300 instances provide 8xNVIDIA Blackwell Ultra GPUs with 2.1 TB high bandwidth GPU memory, 6.4 Tbps EFA networking, 300 Gbps dedicated ENA throughput, and 4 TB of system memory.
P6-B300 instances deliver 2x networking bandwidth, 1.5x GPU memory size, and 1.5x GPU TFLOPS (at FP4, without sparsity) compared to P6-B200 instances, making them well suited to train and deploy large trillion-parameter foundation models (FMs) including large language models (LLMs) with sophisticated techniques. The higher networking and larger memory deliver faster training times and more token throughput for AI workloads.
P6-B300 instances are now available in p6-b300.48xlarge size in the following AWS Regions: US West (Oregon), AWS GovCloud (US-East), US East (N. Virginia), Asia Pacific (Hyderabad, Jakarta, Seoul), and South America (Sao Paulo) Regions. To learn more about P6-B300 instances, visit Amazon EC2 P6 instances.
Starting today, Amazon Elastic Cloud Compute (Amazon EC2) P6-B300 instances are available in the AWS Asia Pacific (Jakarta) Region. P6-B300 instances provide 8xNVIDIA Blackwell Ultra GPUs with 2.1 TB high bandwidth GPU memory, 6.4 Tbps EFA networking, 300 Gbps dedicated ENA throughput, and 4 TB of system memory.
P6-B300 instances deliver 2x networking bandwidth, 1.5x GPU memory size, and 1.5x GPU TFLOPS (at FP4, without sparsity) compared to P6-B200 instances, making them well suited to train and deploy large trillion-parameter foundation models (FMs) including large language models (LLMs) with sophisticated techniques. The higher networking and larger memory deliver faster training times and more token throughput for AI workloads.
P6-B300 instances are now available in p6-b300.48xlarge size in the following AWS Regions: US West (Oregon), AWS GovCloud (US-East), US East (N. Virginia), Asia Pacific (Hyderabad, Jakarta, Seoul), and South America (Sao Paulo) Regions. To learn more about P6-B300 instances, visit Amazon EC2 P6 instances.
Microsoft lleva a ANDICOM 2026 una visión de la nueva frontera de la IA: personas y agentes redefiniendo cómo se organiza el trabajo
El 63% de los usuarios de IA en Colombia afirma que hoy realiza trabajos que no habría podido hacer hace un año.
El siguiente desafío será convertir esa capacidad individual en una nueva forma de operar, combinando personas y agentes sobre datos confiables, contexto de negocio y gobierno.
Cartagena, Colombia – La inteligencia artificial (IA) está entrando en una nueva etapa. Después de incorporarse al trabajo para ayudar a crear, analizar y resumir información, comienza a asumir tareas y participar de manera más activa en procesos. Este cambio plantea un desafío para las organizaciones: pasar de utilizar IA en actividades aisladas a rediseñar cómo se distribuye el trabajo entre personas, agentes, datos y sistemas.
Esta es una de las principales ideas que Microsoft propuso en ANDICOM 2026, cuyo lema, Unleashing the Power of AI, puso el foco en cómo aprovechar y escalar el potencial de esta tecnología. Para la compañía, la próxima fase de adopción estará marcada por la capacidad de las organizaciones para transformar procesos, roles y decisiones alrededor de una inteligencia cada vez más disponible.
Las señales de este cambio ya son visibles entre los trabajadores colombianos. De acuerdo con el Microsoft Work Trend Index 2026, el 63% de los usuarios de IA en el país afirma que actualmente realiza trabajos que no habría podido hacer un año atrás. Sin embargo, esa expansión de capacidades convive con una tensión: el 70% teme quedarse atrás si no adopta la tecnología rápidamente, mientras que el 46% considera más seguro concentrarse en los objetivos actuales de su empresa que rediseñar su manera de trabajar.
“La próxima frontera de la inteligencia artificial es la colaboración entre personas y agentes. Cuando la creatividad, el juicio y la experiencia humana se combinan con agentes capaces de actuar sobre datos y procesos, se crean nuevas oportunidades para innovar, acelerar la toma de decisiones y ampliar lo que las organizaciones pueden lograr. Las empresas que liderarán esta transformación serán aquellas que conecten personas, agentes y datos de forma responsable, convirtiendo el potencial de la IA en una ventaja competitiva real”, afirmó Marc Reguera, Worldwide General Manager Customer Engagement para Microsoft Fabric.
En esta visión, la Empresa Frontera representa un modelo organizacional en el que la IA se integra de manera más profunda a la operación, con inteligencia disponible bajo demanda y equipos formados por personas y agentes. Su lógica es la de organizaciones operadas con IA y lideradas por personas, donde la tecnología amplía la capacidad de analizar y ejecutar, mientras las personas mantienen la dirección, el criterio y la responsabilidad.
Este modelo también modifica la relación cotidiana de las personas con la tecnología. Dependiendo de la tarea, un trabajador puede actuar como autor, utilizando la IA como apoyo; como editor, revisando una propuesta generada por ella; como director, definiendo un objetivo para que un agente lo ejecute; o como orquestador, coordinando un flujo en el que intervienen distintos agentes. La clave está en determinar deliberadamente qué papel debe desempeñar cada parte.
Convertir la ambición de la IA en una ventaja empresarial medible
Llevar este modelo a escala exige una base tecnológica capaz de combinar dos elementos inseparables: datos confiables y contexto de negocio. Los datos proporcionan una base segura, actualizada y gobernada sobre la cual puede operar la IA; el contexto les da significado al incorporar las métricas, relaciones, reglas y procesos propios de cada organización. Ambos son necesarios para que la IA pueda operar de manera útil y confiable a escala.
Sobre esta base, Microsoft IQ articula las distintas dimensiones del contexto que necesita la IA para comprender y actuar dentro de una organización. Work IQ aporta conocimiento sobre cómo trabajan las personas; Fabric IQ, sobre cómo opera el negocio; y Foundry IQ, sobre el conocimiento que utilizan los agentes. A esto se suma Agent 365, que incorpora capacidades de observabilidad, administración y gobierno de agentes. En conjunto, esta arquitectura reúne personas, datos y agentes bajo un contexto compartido y gobernado para integrar la IA de manera más profunda a la operación.
Esta evolución también está ampliando quién puede construir tecnología dentro de una empresa. Con agentes capaces de traducir una intención de negocio en código funcional, la frontera entre el trabajador de información y el desarrollador comienza a hacerse más flexible. Un usuario puede iniciar una aplicación y un desarrollador profesional continuar posteriormente sobre la misma base tecnológica. El resultado es una nueva distribución de capacidades.
Liberar el verdadero potencial de la IA requiere conectar personas, agentes, datos, contexto y gobierno alrededor de resultados concretos. El propósito para las organizaciones debe ser integrar la IA de manera deliberada para ampliar capacidades, enriquecer la experiencia de empleados y clientes, rediseñar procesos de negocio y acelerar la innovación. El siguiente salto de la IA no ocurrirá únicamente en las herramientas que utilizan las personas, sino en la capacidad de las organizaciones para transformar la manera en que trabajan, deciden y crean valor.
Acerca de Microsoft
Microsoft (Nasdaq «MSFT» @microsoft) crea plataformas y herramientas impulsadas por la IA para ofrecer soluciones innovadoras que satisfagan las necesidades cambiantes de nuestros clientes. La empresa de tecnología está comprometida con hacer que la IA esté ampliamente disponible, de manera responsable, con la misión de empoderar a cada persona y a cada organización en el planeta para lograr más.
The post Microsoft lleva a ANDICOM 2026 una visión de la nueva frontera de la IA: personas y agentes redefiniendo cómo se organiza el trabajo appeared first on Source LATAM.
AWS Gateway Load Balancer (GWLB) now supports sending TCP Reset (RST) packets when a target becomes unhealthy, is deregistered, or when a flow’s idle timeout expires. This feature helps reduce traffic interruptions from minutes to seconds by enabling TCP endpoints to quickly detect failed connections and establish new TCP flows through healthy targets.
Previously, when a GWLB target failed, existing TCP connections would continue to be forwarded to the unhealthy target (aka fail-open behavior). Client and server applications could experience interruptions lasting several minutes due to TCP retry and exponential back-off mechanisms built in the TCP stacks of clients or servers. When this capability is enabled, GWLB sends TCP Resets in response to the incoming traffic, indicating to the sender that a TCP connection is no longer viable. This allows TCP endpoints to recover in quickly.
TCP Reset is not enabled by default to ensure backward compatibility. You can enable it per target group using the AWS Management Console, AWS CLI, or API. Three triggers are supported independently for generating TCP Reset: target becomes unhealthy, target deregistration (after connection draining), and TCP idle timeout expiry.
This feature is available for all new and existing Gateway Load Balancers in all AWS Regions where GWLB is available. There is no additional charge for using this feature.
To learn more, visit this AWS blog,and GWLB User Guide here and here.
AWS Gateway Load Balancer (GWLB) now supports sending TCP Reset (RST) packets when a target becomes unhealthy, is deregistered, or when a flow’s idle timeout expires. This feature helps reduce traffic interruptions from minutes to seconds by enabling TCP endpoints to quickly detect failed connections and establish new TCP flows through healthy targets. Previously, when a GWLB target failed, existing TCP connections would continue to be forwarded to the unhealthy target (aka fail-open behavior). Client and server applications could experience interruptions lasting several minutes due to TCP retry and exponential back-off mechanisms built in the TCP stacks of clients or servers. When this capability is enabled, GWLB sends TCP Resets in response to the incoming traffic, indicating to the sender that a TCP connection is no longer viable. This allows TCP endpoints to recover in quickly. TCP Reset is not enabled by default to ensure backward compatibility. You can enable it per target group using the AWS Management Console, AWS CLI, or API. Three triggers are supported independently for generating TCP Reset: target becomes unhealthy, target deregistration (after connection draining), and TCP idle timeout expiry. This feature is available for all new and existing Gateway Load Balancers in all AWS Regions where GWLB is available. There is no additional charge for using this feature. To learn more, visit this AWS blog, and GWLB User Guide here and here.
