AWS wellarchitected documentation change
Summary
Updated prescriptive guidance for energy-efficient workload scheduling, added recommendations for Amazon SQS, generative AI response balancing, cost-aware prompting, and distributed inference.
Security assessment
Changes focus on optimizing computational workloads and generative AI for energy efficiency and resource utilization. No security vulnerabilities, incidents, or security controls are mentioned. The added fault tolerance reference relates to operational reliability, not security.
Diff
diff --git a/wellarchitected/latest/financial-services-industry-lens/fsisus13.md b/wellarchitected/latest/financial-services-industry-lens/fsisus13.md index 9cbf79e97..bd4cc6a48 100644 --- a//wellarchitected/latest/financial-services-industry-lens/fsisus13.md +++ b//wellarchitected/latest/financial-services-industry-lens/fsisus13.md @@ -21 +21 @@ If your workload does not have time-sensitive requirements, consider running the -**Prescriptive guidance** +### Prescriptive guidance @@ -25 +25,7 @@ If your workload does not have time-sensitive requirements, consider running the - * **Implementation guidance:** Evaluate whether highly intensive computational workloads such as financial simulation can be spread over time and run fewer instances to maximize renewable energy availability. If a grid computing workload is using a third-party scheduler, prioritize workloads that need to provide calculations for regulators and trading desks that need information prior to markets opening, so workloads that are not urgent can be pushed off and worked on at a consistent rate to maximize renewable energy availability. Additionally, verify that a proper fault tolerance framework is implemented, as restarting a launch can increase launch time and energy consumption. + * Evaluate whether highly intensive computational workloads such as financial simulation can be spread over time and run fewer instances to maximize renewable energy availability. If a grid computing workload is using a third-party scheduler, prioritize workloads that need to provide calculations for regulators and trading desks that need information prior to markets opening, so workloads that are not urgent can be pushed off and worked on at a consistent rate to maximize renewable energy availability. Additionally, verify that a proper fault tolerance framework is implemented, as restarting a launch can increase launch time and energy consumption. + + * Use [Amazon Simple Queue Service (Amazon SQS)](https://aws.amazon.com/sqs/) achieve your goal. + + * Balance generative AI model response time requirements with energy efficiency. + + * Implement cost-aware prompting strategies that may take slightly longer but use fewer resources. @@ -27 +33 @@ If your workload does not have time-sensitive requirements, consider running the - * **Service recommendations:** Use [Amazon Simple Queue Service (Amazon SQS)](https://aws.amazon.com/sqs/) achieve your goal. + * Use distributed generative AI inference when time permits to optimize resource utilization.