Global IAM identity scope primarily supports the earlier need to manage IAM identities used to authorize actions across AWS Regions because IAM is a global AWS service within an account, while service resources may be regional, but the new task is to route user traffic toward healthy regional endpoints through AWS global network paths; that is why AWS Global Accelerator is preferable here.

Availability Zone is intended to separate redundant application components across isolated facilities within one Region because an Availability Zone is one or more discrete data centers with independent infrastructure in a Region; in this design the new outcome is to route user traffic toward healthy regional endpoints through AWS global network paths, which calls for AWS Global Accelerator rather than Availability Zone or the earlier Global IAM identity scope.

AWS Region is intended to select a geographic area in which to place a workload because an AWS Region is a distinct geographic location containing multiple Availability Zones; in this design the new outcome is to route user traffic toward healthy regional endpoints through AWS global network paths, which calls for AWS Global Accelerator rather than AWS Region or the earlier Global IAM identity scope.

That description belongs to Accelerated computing EC2 instance because accelerated instance families provide processors suited to parallel graphics or ML workloads; it is used to handle workloads requiring GPUs or specialized hardware acceleration, not to explain how General purpose EC2 instance provides the function to balance CPU memory and networking for broadly mixed workloads.

That description belongs to Memory optimized EC2 instance because memory-optimized families provide higher RAM-to-vCPU ratios for memory-bound applications; it is used to support applications with large in-memory working sets, not to explain how General purpose EC2 instance provides the function to balance CPU memory and networking for broadly mixed workloads.

That description belongs to Compute optimized EC2 instance because compute-optimized instance families emphasize processing power for compute-bound tasks; it is used to provide comparatively more CPU resources for CPU-intensive applications, not to explain how General purpose EC2 instance provides the function to balance CPU memory and networking for broadly mixed workloads.

That description belongs to Private subnet because a private subnet can reach only the networks available through its route table; it is used to place backend resources without a direct internet gateway route, not to explain how Amazon API Gateway provides the function to expose and manage HTTP REST or WebSocket APIs.

That description belongs to Public subnet because a public subnet has a route to an attached internet gateway where applicable; it is used to place resources in a subnet with a route toward an internet gateway, not to explain how Amazon API Gateway provides the function to expose and manage HTTP REST or WebSocket APIs.

That description belongs to Amazon VPC because a VPC defines virtual networking boundaries and hosts configurable subnets and routing; it is used to create a logically isolated IP network for AWS resources, not to explain how Amazon API Gateway provides the function to expose and manage HTTP REST or WebSocket APIs.

Database engine choice by use case addresses the separate need to match persistence technology to workload queries and consistency needs because AWS purpose-built databases span relational key-value document graph and in-memory patterns; it is not as direct as ACID relational transactions for the actual requirement to preserve multi-step data consistency in a relational workload.

RDS read replica addresses the separate need to offload eligible read queries from a relational database primary because read replicas provide copies usable to serve read traffic with replication behavior; it is not as direct as ACID relational transactions for the actual requirement to preserve multi-step data consistency in a relational workload.

DynamoDB global tables addresses the separate need to provide multi-Region active-active replication for a DynamoDB application because global tables replicate supported data across selected Regions for regional workloads; it is not as direct as ACID relational transactions for the actual requirement to preserve multi-step data consistency in a relational workload.

DynamoDB point-in-time recovery addresses the separate need to restore supported table data to a recent earlier point because pITR continuously backs up eligible DynamoDB table data for point-in-time restore within its retention window; it is not as direct as ACID relational transactions for the actual requirement to preserve multi-step data consistency in a relational workload.

AWS Migration Hub addresses the separate need to track multiple migration activities in one view because migration Hub centralizes migration progress information from supported tools; it is not as direct as ACID relational transactions for the actual requirement to preserve multi-step data consistency in a relational workload.

Elastic Load Balancing is intended to spread inbound traffic across multiple eligible targets because eLB directs client traffic toward configured healthy application targets; in this design the new outcome is to handle high-performance connection-level TCP or UDP traffic, which calls for Network Load Balancer rather than Elastic Load Balancing or the earlier Gateway Load Balancer.

Elasticity through scaling policies is intended to match compute capacity to traffic fluctuations automatically because metric-driven policies can change running capacity in response to utilization or demand; in this design the new outcome is to handle high-performance connection-level TCP or UDP traffic, which calls for Network Load Balancer rather than Elasticity through scaling policies or the earlier Gateway Load Balancer.

Gateway Load Balancer primarily supports the earlier need to insert compatible virtual network appliances into traffic paths because gateway Load Balancer distributes flows across appliance targets for inspection or similar tasks, but the new task is to handle high-performance connection-level TCP or UDP traffic; that is why Network Load Balancer is preferable here.

It differs from training a classical model entirely from raw data with a custom pipeline; that caveat is associated more closely with Amazon Bedrock for the need to build generative AI applications using foundation models through managed APIs than with Amazon Lex for the requirement to build conversational chat or voice interfaces with intent recognition.

It provides customization beyond consuming a fixed pretrained AI API; that caveat is associated more closely with Amazon SageMaker AI for the need to build train and deploy custom machine learning models than with Amazon Lex for the requirement to build conversational chat or voice interfaces with intent recognition.

It does not primarily convert speech waveforms into written transcripts; that caveat is associated more closely with Amazon Comprehend for the need to extract sentiment entities and other insights from text than with Amazon Lex for the requirement to build conversational chat or voice interfaces with intent recognition.

That description belongs to S3 Same-Region Replication because same-Region Replication can support same-region copy and ownership workflows; it is used to replicate eligible objects between buckets within the same AWS Region, not to explain how Amazon S3 One Zone-IA provides the function to store recreatable infrequently accessed objects within one Availability Zone.

That description belongs to Amazon S3 Standard because S3 Standard is appropriate for data requiring regular access across multiple AZs; it is used to serve frequently accessed objects with low-latency retrieval, not to explain how Amazon S3 One Zone-IA provides the function to store recreatable infrequently accessed objects within one Availability Zone.

That description belongs to Amazon S3 because S3 is an object storage service for files and data blobs rather than mounted block devices; it is used to store durable objects and retrieve them by bucket and key, not to explain how Amazon S3 One Zone-IA provides the function to store recreatable infrequently accessed objects within one Availability Zone.

Cross-Region replication primarily supports the earlier need to maintain copies of selected data outside its source Region because cross-Region replication moves eligible objects or database data into a different geography when configured, but the new task is to recognize that many AWS resources exist only in a selected Region; that is why Regional service scope is preferable here.

Data residency by Region is intended to keep stored customer data in a selected geography because region placement and configured replication determine where supported AWS workload data resides; in this design the new outcome is to recognize that many AWS resources exist only in a selected Region, which calls for Regional service scope rather than Data residency by Region or the earlier Cross-Region replication.

Regional service availability check is intended to verify a particular offering is available in the chosen Region because AWS service availability varies by Region and sometimes by instance family or feature; in this design the new outcome is to recognize that many AWS resources exist only in a selected Region, which calls for Regional service scope rather than Regional service availability check or the earlier Cross-Region replication.

Amazon Route 53 is intended to resolve application domain names and apply supported DNS routing policies because Route 53 is a managed DNS and domain service with traffic-routing and health-check features; in this design the new outcome is to distribute DNS responses among configured endpoints according to weights, which calls for Route 53 weighted routing rather than Amazon Route 53 or the earlier Route 53 latency routing.

Route 53 geolocation routing is intended to select DNS answers based on the geographic origin of DNS queries because geolocation rules can direct supported query origins to mapped endpoints; in this design the new outcome is to distribute DNS responses among configured endpoints according to weights, which calls for Route 53 weighted routing rather than Route 53 geolocation routing or the earlier Route 53 latency routing.

Route 53 latency routing primarily supports the earlier need to answer DNS requests with an endpoint selected for lower network latency because latency-based policies choose among configured Region records using AWS latency observations, but the new task is to distribute DNS responses among configured endpoints according to weights; that is why Route 53 weighted routing is preferable here.

Region outage planning primarily supports the earlier need to prepare recovery beyond failure of one Availability Zone because a regional recovery design uses another Region and explicit data and traffic recovery plans, but the new task is to address disaster recovery or geography-spanning application availability; that is why Multi-Region architecture is preferable here.

Data residency by Region is intended to keep stored customer data in a selected geography because region placement and configured replication determine where supported AWS workload data resides; in this design the new outcome is to address disaster recovery or geography-spanning application availability, which calls for Multi-Region architecture rather than Data residency by Region or the earlier Region outage planning.

Multi-AZ application deployment is intended to survive a failure confined to one Availability Zone because replicated tiers across separate AZs reduce dependence on a single facility group; in this design the new outcome is to address disaster recovery or geography-spanning application availability, which calls for Multi-Region architecture rather than Multi-AZ application deployment or the earlier Region outage planning.

NoSQL access-pattern design primarily supports the earlier need to design a database around known key lookups and throughput patterns because DynamoDB data modeling focuses on partition keys indexes and predictable access paths, but the new task is to restore supported table data to a recent earlier point; that is why DynamoDB point-in-time recovery is preferable here.

Amazon Redshift is intended to analyze large structured data sets in a columnar warehouse because redshift provides a cloud data warehouse optimized for analytical SQL workloads; in this design the new outcome is to restore supported table data to a recent earlier point, which calls for DynamoDB point-in-time recovery rather than Amazon Redshift or the earlier NoSQL access-pattern design.

Amazon ElastiCache is intended to cache frequently accessed application data in memory because elastiCache offers managed in-memory caching engines to reduce repetitive backend reads; in this design the new outcome is to restore supported table data to a recent earlier point, which calls for DynamoDB point-in-time recovery rather than Amazon ElastiCache or the earlier NoSQL access-pattern design.