AWS wellarchitected documentation change
Summary
Restructured prescriptive guidance for network optimization, removed problem statement header, added generative AI inference optimizations and prompt engineering techniques
Security assessment
Changes emphasize network efficiency and sustainability (data transfer reduction, scalable services) without security references. Added AI optimizations focus on performance/token efficiency, not security. 'Risky configurations' refers to sustainability inefficiencies, not security vulnerabilities.
Diff
diff --git a/wellarchitected/latest/financial-services-industry-lens/fsisus05.md b/wellarchitected/latest/financial-services-industry-lens/fsisus05.md index 58d5351c1..9b47b058b 100644 --- a//wellarchitected/latest/financial-services-industry-lens/fsisus05.md +++ b//wellarchitected/latest/financial-services-industry-lens/fsisus05.md @@ -19,2 +18,0 @@ Remove redundant layers and redirects, use pagination and local caching mechanis -**Problem statement** - @@ -25 +23 @@ A simple example of this is using [AWS Direct Connect](https://docs.aws.amazon.c -Another common mistake is to serve both OLAP and OLTP workloads from the same database or cluster, which normally span two or more completely different geographic locations. Both of these patterns generate excessive and unnecessary data movement. +Another common mistake is to serve both OLAP and OLTP workloads from the same database or cluster, which normally span two or more completely different geographic locations. Both patterns generate excessive and unnecessary data movement. @@ -27 +25,11 @@ Another common mistake is to serve both OLAP and OLTP workloads from the same da -**Prescriptive guidance** +### Prescriptive guidance + +Identify poor architectural choices and risky configurations as good candidates for remediation. + +Assess your workflows from the perspective of varying demand over time, so select scalable AWS services over fixed ones. + +Do not underestimate your network requirements, especially for peak loads. Provide sufficient failover resources to support your operations in case of partial outages. + +Optimize generative AI inference patterns to minimize data transfer and network overhead. + +Implement edge inference for generative AI models where appropriate to reduce network traffic. @@ -29 +37 @@ Another common mistake is to serve both OLAP and OLTP workloads from the same da -Identify poor architectural choices and risky configurations as good candidates for remediation. Assess your workflows from the perspective of varying demand over time, so select scalable AWS services over fixed ones. Do not underestimate your network requirements, especially for peak loads. Provide sufficient failover resources to support your operations in case of partial outages. +Use efficient prompt engineering to reduce token lengths and network utilization.