AWS sagemaker documentation change
Summary
Updated documentation for SageMaker training plans to include Inference endpoints and Studio apps as target resources, reorganized content flow, added new instance types, and clarified reservation policies
Security assessment
The changes primarily expand supported resources (Inference endpoints/Studio apps), add instance types, and reorganize content. No security vulnerabilities, patches, or security-specific features are mentioned. Modifications focus on product capabilities and reservation workflows without addressing security weaknesses.
Diff
diff --git a/sagemaker/latest/dg/reserve-capacity-with-training-plans.md b/sagemaker/latest/dg/reserve-capacity-with-training-plans.md index cd063a9a0..5525fab4f 100644 --- a//sagemaker/latest/dg/reserve-capacity-with-training-plans.md +++ b//sagemaker/latest/dg/reserve-capacity-with-training-plans.md @@ -7 +7 @@ -What are SageMaker training plansBenefitsReservationUser workflowSupported instance types, AWS Regions, and pricingUltraServers in SageMaker AISearch behaviorConsiderations +What are SageMaker training plansBenefitsReservationSupported instance types, AWS Regions, and pricingUltraServers in SageMaker AISearch behaviorConsiderationsUser workflow @@ -9 +9 @@ What are SageMaker training plansBenefitsReservationUser workflowSupported insta -# Reserve training plans for your training jobs or HyperPod clusters +# Reserve Flexible Training Plans for ML workloads @@ -11 +11 @@ What are SageMaker training plansBenefitsReservationUser workflowSupported insta -Amazon SageMaker training plans is a capability that allows you to reserve and help maximize the use of GPU capacity for large-scale AI model training workloads. This feature provides access to highly sought-after instance types that cover a range of GPU-accelerated computing options, including the latest NVIDIA GPU technologies and AWS trainium chips. With SageMaker training plans, you can secure predictable access to these high-demand, high-performance computational resources within your specified timelines and budgets, without the need to manage underlying infrastructure. This flexibility is particularly valuable for organizations dealing with the challenges of acquiring and scheduling these oversubscribed compute instances for their mission-critical AI workloads. +Amazon SageMaker training plans is a capability that allows you to reserve and help maximize the use of GPU capacity for ML workloads. This feature provides access to highly sought-after instance types that cover a range of GPU-accelerated computing options, including the latest NVIDIA GPU technologies and AWS Trainium chips. With SageMaker training plans, you can secure predictable access to these high-demand, high-performance computational resources within your specified timelines and budgets, without the need to manage underlying infrastructure. This flexibility is particularly valuable for organizations dealing with the challenges of acquiring and scheduling these oversubscribed compute instances for their mission-critical AI workloads. @@ -15 +15 @@ Amazon SageMaker training plans is a capability that allows you to reserve and h -SageMaker training plans allow you to reserve compute capacity tailored to your target resource needs, such as SageMaker training jobs or SageMaker HyperPod clusters. The service automatically handles the reservation, provisioning of accelerated compute resources, infrastructure setup, workload execution, and recovery from infrastructure failures. +SageMaker training plans allow you to reserve compute capacity tailored to your target resource needs, such as SageMaker training jobs, SageMaker HyperPod clusters, SageMaker Inference endpoints, or SageMaker Studio apps. The service automatically handles the reservation, provisioning of accelerated compute resources, infrastructure setup, workload execution, and recovery from infrastructure failures. @@ -34 +34 @@ SageMaker training plans consist of one or more Reserved Capacity blocks, each d - * Training plans are specific to their target resource (either SageMaker Training Job or SageMaker HyperPod) and cannot be interchanged. + * Training plans are specific to their target resource (SageMaker Training Job, SageMaker HyperPod, SageMaker Inference endpoints, or SageMaker Studio apps) and cannot be interchanged. @@ -47 +47 @@ SageMaker training plans offer the following benefits: - * **Cost Management** : Plan and budget for large-scale training requirements in advance. + * **Cost Management** : Plan and budget for your ML workload requirements in advance. @@ -51 +51 @@ SageMaker training plans offer the following benefits: - * **Flexibility** : Create training plans for various resources, including SageMaker training jobs and SageMaker HyperPod clusters. + * **Flexibility** : Create training plans for various resources, including SageMaker training jobs, SageMaker HyperPod clusters, SageMaker Inference endpoints, and SageMaker Studio apps. @@ -53 +53 @@ SageMaker training plans offer the following benefits: - * **Fault Tolerance** : Benefit from automatic recovery from infrastructure failures and workload migration across Availability Zones for SageMaker AI training jobs. + * **Fault Tolerance** : Benefit from automatic recovery from infrastructure failures and workload migration across Availability Zones for SageMaker training jobs. @@ -72,4 +71,0 @@ You can search for and purchase a plan that will be accessible within 30 minutes - * **UltraServers** : Reach out to your account manager to request UltraServers. - - * Add **P5.4xl, P4de, B300, G6** : Reach out to your account manager to request these instance types. - @@ -78 +74 @@ You can search for and purchase a plan that will be accessible within 30 minutes - * **Training plan termination** : If you're using training jobs as a target resource and 30 minutes remain in a Reserved Capacity, SageMaker training plans initiates the process of terminating any running instances within that block until the next Reserved Capacity becomes active. You retain full access to your training plan until 30 minutes before the final Reserved Capacity block's end time. + * **Training plan termination** : If you're using Training Jobs, Inference endpoints, and Studio apps as a target resource and 30 minutes remain in a Reserved Capacity, SageMaker training plans initiates the process of terminating any running instances within that block until the next Reserved Capacity becomes active. You retain full access to your training plan until 30 minutes before the final Reserved Capacity block's end time. @@ -85,41 +80,0 @@ If your target resource is a SageMaker HyperPod cluster, this time limit is one -## SageMaker training plans user workflow - -SageMaker training plans work through the following steps: - -Admin steps: - - 1. **Search and review** : Find available plan offerings that match your compute requirements, such as instance type, count, start time, and duration. - - 2. **Create a plan** : Reserve a training plan that meets your needs using the ID of your chosen plan offering. - - 3. **Payment and scheduling** : Upon successful upfront payment, the plan status becomes `Scheduled`. - - - - -Steps for plan users / ML engineers: - - 1. **Resource allocation** : Use your plan to queue SageMaker AI training jobs or allocate to a SageMaker HyperPod cluster instance group. - - 2. **Activation** : When the plan start date arrives, it becomes `Active`. Based on available reserved capacity, SageMaker training plans automatically launch training jobs or provision instance groups. - - - - -###### Note - -The status of the training plan transitions from `Scheduled` to `Active` when a Reserved Capacity period begins, and then back to `Scheduled` when waiting for the next Reserved Capacity period to start. - -The following diagrams provide a comprehensive overview of how SageMaker training plans interact with different target resources, illustrating a plan's lifecycle and its role in resource allocation for both SageMaker training jobs and SageMaker HyperPod clusters. - - * **Training plans for SageMaker Training Job** : The first diagram illustrates the end-to-end workflow of the interaction between a training plan and SageMaker Training Job. - - - - * **Training plans for SageMaker HyperPod clusters** : The second diagram illustrates the end-to-end workflow of the interaction between a training plan and a SageMaker HyperPod instance group. - - - - - - @@ -131,0 +87,2 @@ Training plans support reservations for the following specific high-performance + * **ml.p5.4xlarge** + @@ -144 +101 @@ Training plans support reservations for the following specific high-performance - * **ml.c6i-32xlargesc** + * **ml.p6-b300.48xlarge** @@ -145,0 +103 @@ Training plans support reservations for the following specific high-performance + * **ml.g6.xlarge** (reach out to your account manager) @@ -146,0 +105 @@ Training plans support reservations for the following specific high-performance + * **ml.g6.4xlarge** (reach out to your account manager) @@ -149 +108,3 @@ Training plans support reservations for the following specific high-performance -**UltraServers** + + +**UltraServers** (Currently cannot be purchased in a self-serve manner; reach out to your account manager.) @@ -227 +188,3 @@ SageMaker training plans return up to three offerings of one or two segments. Fo - * Training plans cannot be modified once purchased. + * Training plan purchases are final and cannot be cancelled. + + * Training plans cannot be modified to add or remove instances; they can only be extended to a new end date. @@ -256,0 +220,41 @@ This multi-AZ approach for training jobs provides greater flexibility in resourc +## SageMaker training plans user workflow + +SageMaker training plans work through the following steps: + +Admin steps: + + 1. **Search and review** : Find available plan offerings that match your compute requirements, such as instance type, count, start time, and duration. + + 2. **Create a plan** : Reserve a training plan that meets your needs using the ID of your chosen plan offering. + + 3. **Payment and scheduling** : Upon successful upfront payment, the plan status becomes `Scheduled`. + + + + +Steps for plan users / ML engineers: + + 1. **Resource allocation** : Use your plan to allocate to SageMaker AI training jobs, SageMaker HyperPod cluster instance groups, SageMaker Inference endpoints, or spaces in SageMaker Studio apps. + + 2. **Activation** : When the plan start date arrives, it becomes `Active`. Based on available reserved capacity, SageMaker training plans automatically provision training jobs, instance groups, inference endpoints, or Studio applications. + + + + +###### Note + +The status of the training plan transitions from `Scheduled` to `Active` when a Reserved Capacity period begins, and then back to `Scheduled` when waiting for the next Reserved Capacity period to start. + +The following diagrams provide a comprehensive overview of how SageMaker training plans interact with different target resources, illustrating a plan's lifecycle and its role in resource allocation for both SageMaker training jobs and SageMaker HyperPod clusters. + + * **Training plans for SageMaker Training Job** : The first diagram illustrates the end-to-end workflow of the interaction between a training plan and SageMaker Training Job. + + + + * **Training plans for SageMaker HyperPod clusters** : The second diagram illustrates the end-to-end workflow of the interaction between a training plan and a SageMaker HyperPod instance group. + + + + + +