This question centers on improving scalability and global access performance.

Auto Scaling groups enable EC2 instances to scale dynamically in response to demand, ensuring availability during peak hours without manual intervention. Amazon RDS read replicas offload read traffic, improving read throughput and reducing latency on the primary database instance. Deploying Amazon CloudFront with S3 as origin accelerates delivery of static transaction documents globally by caching content at edge locations, reducing latency for users worldwide.

Option B focuses on vertical scaling (larger instances) and caching with ElastiCache, but it does not address global content delivery optimally. AWS Global Accelerator accelerates network traffic but is better suited for accelerating TCP and UDP traffic; CloudFront is generally preferred for HTTP content delivery.

Option C migrates workloads to Lambda and Aurora global databases, which is an advanced and potentially costly redesign that may not be necessary. Option D suggests moving to ECS and multi-AZ RDS but does not address global content delivery efficiently.

Therefore, option A uses proven scalability and caching best practices aligned with AWS Well-Architected Framework pillars for performance and operational excellence.

AWS Step Functions supports long-running workflows and error retries, making it ideal for a job composed of many tasks. Integration with EventBridge allows automatic triggering from S3 events. This setup is resilient and supports up to 1-year execution duration.

AWS Lambda supports functions up to 10 GB of memory and 15 minutes execution time. Each invocation here requires only 1 GB of memory and finishes in 30 seconds, making it an ideal fit for Lambda. Lambda is cost-effective for event-driven, short-duration workloads and requires no infrastructure management. AWS Glue and EMR are better suited for large-scale ETL or distributed processing, which is unnecessary and more costly for this workload.

AWS Documentation Extract:

“AWS Lambda is a serverless compute service that lets you run code without provisioning or managing servers. You pay only for the compute time you consume. Lambda supports up to 10 GB memory and 15 minutes per invocation.”

(Source: AWS Lambda documentation)

B, D: AWS Glue is intended for ETL on larger datasets or batch jobs, usually with higher operational overhead and cost.

C: EMR is for large-scale distributed processing and is not cost-effective for single, fast, memory-bound jobs.

AWS Budgets allows you to set custom cost and usage budgets. When actual or forecasted usage exceeds the threshold, AWS Budgets can automatically send alerts to an Amazon SNS topic. You can use AWS Cost Explorer in parallel for visual tracking of spending. This solution requires no code and has minimal operational overhead.

AWS Lambda is not a good fit here because AWS documents that a standard Lambda invocation can run forup to 15 minutes, while this workload can takeup to 30 minutes. That makes the Lambda-based answers unsuitable. AnSQS standard queueis a good match for decoupling uploads from processing and for allowing multiple jobs to run in parallel. Running the processors onAmazon ECS with AWS Fargateminimizes operational overhead because the company does not have to manage EC2 hosts. ECS service scaling can also be driven from queue-backed metrics, allowing the processing fleet to expand as the queue grows. Therefore, S3 notifications to SQS with Fargate-based processing is the most appropriate low-overhead design.

============

AWS Secrets Manager is purpose-built to securely store, manage, and rotate sensitive credentials such as database user names and passwords. Option A is the most operationally efficient solution because it eliminates manual password rotation, reduces human error, and centralizes secret lifecycle management. Secrets Manager integrates natively with Amazon Aurora, enabling automated credential rotation using AWS-managed or custom Lambda rotation logic. Once rotation is enabled, Secrets Manager updates the database credentials and stores the new values securely without requiring administrators to manually update files on EC2 instances.

By assigning IAM permissions to the secret, access can be tightly controlled using least-privilege principles. The application retrieves credentials at runtime, removing the need to store passwords locally on disk, which significantly improves security posture. Secrets Manager also provides auditing capabilities through AWS CloudTrail, allowing visibility into secret access and changes.

Option B (Systems Manager Parameter Store) can securely store secrets, but automated rotation is not natively supported in the same way as Secrets Manager. Implementing rotation with Parameter Store would require additional custom automation, increasing operational complexity. Option C stores credentials in S3, which is not designed for frequent credential rotation or secure secret access patterns, even when encrypted. Option D only encrypts credentials at rest on individual instances and does not address rotation, distribution, or centralized management, resulting in high operational overhead.

Therefore, A best meets the requirements by providing secure storage, automated monthly rotation, fine-grained access control, and minimal operational effort, aligning with AWS security and operational excellence best practices.

AWS Secrets Manager is the managed service for securely storing, rotating, and retrieving database credentials. When you store Aurora credentials in Secrets Manager and enable automatic rotation, Secrets Manager updates the credentials in both the database and the stored secret.

Each Lambda function version or alias can call Secrets Manager at runtime to retrieve the current secret value, so even older deployed Lambda versions that use an alias will always obtain the most recent credentials without redeployment.

Why others are not suitable:

B: Embeds credentials in code, requiring redeployment on every rotation and violating security best practices.

C: Environment variables are version-specific; old aliases would continue using outdated values unless you redeploy or change them.

D: Parameter Store can store and rotate secrets but is less integrated for database credential rotation than Secrets Manager; Secrets Manager is the purpose-built minimal-overhead choice here.

Amazon CloudFront uses a globally distributed network of edge locations that cache content close to users. When Outposts servers around the world download large software packages, CloudFront provides significantly reduced latency due to edge caching. This is the AWS-recommended solution for accelerating downloads of static files at scale and across global locations.

S3 Transfer Acceleration optimizes uploads and downloads to a single Region but does not provide edge caching. Multi-Region replication with failover does not reduce latency globally because requests still must reach the regional origins. Therefore, CloudFront with caching enabled is the correct design for improving download speed worldwide.

=====================================================

AWS Backup supports cross-Region backup for Amazon EFS, allowing automated, scheduled backups and replication to another Region. This managed service simplifies backup management and ensures data resilience without the need for custom scripts or manual processes

Option A:The ALB must accept HTTPS traffic from the public internet. Allowing inbound traffic on port 443 from 0.0.0.0/0 enables this functionality.

Option C:The ALB must forward HTTPS traffic to the web application servers on port 443. Outbound traffic for port 443 must be allowed for this communication.

Option E:The ALB must perform health checks on the web application servers over HTTPS on port 8443. Outbound traffic for port 8443 must be allowed for this purpose.

Option B:Allowing all outbound traffic is overly permissive and does not align with the specific requirements.

Option D and F:Inbound traffic to the ALB from the web application instances is unnecessary because the flow of traffic is from the ALB to the web application instances, not vice versa.

AWS Documentation References:

Application Load Balancer Security Groups

Health Checks for ALBs

A. API Gateway + DynamoDB:Provides high scalability, low latency, and seamless integration with Amazon Comprehend for NLP tasks.

B. HTTP API + Translate + ElastiCache:Translate is not relevant for NLP tasks like sentiment analysis or entity recognition. ElastiCache is unsuitable for permanent storage.

C. SQS + EC2 + RDS:Increases complexity and operational overhead. RDS may not scale effectively for high concurrent loads.

D. WebSocket API + Textract:Textract is irrelevant for NLP tasks. WebSocket API is not the optimal choice for this use case.

Predictive scaling in EC2 Auto Scaling uses machine learning to analyze historical load and forecast future capacity needs, then pre-scales capacity before the demand occurs. It is specifically designed for recurring, cyclical workloads such as weekly batch jobs.

In this scenario:

The workload pattern is known and recurring (weekly scripted jobs).

The company explicitly does not want to manually analyze trends.

They need capacity to be ready 30 minutes before jobs start.

Predictive scaling lets you:

Set the metric (CPU utilization) and target (60%).

Configure the policy to pre-launch instances 30 minutes in advance, satisfying the requirement automatically with minimal operational overhead.

Dynamic scaling (Option A) reacts after CPU increases, not before.

Scheduled scaling (Option B) requires manual analysis and ongoing adjustment of capacity values.

EventBridge + Lambda (Option D) introduces unnecessary custom logic and still reacts after CPU reaches 60%, not 30 minutes before.

Because the workload follows apredictable business-hours schedule, the lowest-overhead solution is to usescheduled scalingwith EC2 Auto Scaling. AWS documentation states that scheduled actions let you scale an Auto Scaling group at specific times or on recurring schedules by changing desired, minimum, and maximum capacity. That directly matches a workload that needs instances during business hours and not at night. Manual start and stop actions add operational work, Spot Instances risk interruptions during business hours, and sizing for maximum load with Reserved Instances would overpay for idle time. Scheduled scaling automates the exact time-based pattern the question describes.

============

When an AWS Lambda function is invoked, especially after periods of inactivity, it may experience cold starts that delay execution. As demand increases, the scaling behavior and latency of Lambda can affect performance. Provisioned Concurrency is an AWS feature designed specifically to solve this issue.

From AWS Documentation:

“Provisioned Concurrency keeps functions initialized and hyper-ready to respond in double-digit milliseconds.”

(Source: AWS Lambda Developer Guide – Managing Concurrency)

Why Option D is correct:

Provisioned Concurrency ensures that a specified number of Lambda function instances are always warm and ready to serve requests, eliminating cold start latency.

It ' s cost-effective for workloads with consistent usage patterns, like real-time simulations for a small user group.

Maintains scalability and low overhead of Lambda without moving to managed container or EC2 platforms.

Why the other options are less optimal:

Option A (Fargate) and Option B (EC2): Introduce more infrastructure and higher ongoing costs for a small team with likely intermittent usage.

Option C (AWS Batch): Ideal for batch jobs, not real-time simulations; also incurs higher latency due to job queuing.

The issue in this case is related to the processing tier, where EC2 instances are overwhelmed at peak times, causing delays.Option D, using anAmazon EC2 Auto Scaling target tracking policybased on theApproximateNumberOfMessagesin the SQS queue, is the best solution.

Auto Scaling with Target Tracking:

Target tracking policiesdynamically scale out or in based on a specific metric. For this use case, you can monitor theApproximateNumberOfMessagesin the SQS queue. When the number of messages (orders) in the queue increases, the Auto Scaling group will scale out more EC2 instances to handle the additional load, ensuring that the queue doesn’t build up and cause delays.

This solution is ideal for handling variable and unpredictable peak times, as Auto Scaling can automatically adjust based on real-time load rather than scheduled times.

Why Not the Other Options?:

Option A (Scheduled Scaling): Scheduled scaling works well for predictable peak times, but this company experiencesunpredictable peak usage, making scheduled scaling less effective.

Option B (ElastiCache for Redis): Adding a caching layer would help if DynamoDB were the bottleneck, but in this case, the issue is CPU overload on EC2 instances in the processing tier.

Option C (CloudFront): CloudFront would help cache static content from the web tier, but it wouldn’t resolve the issue of the processing tier’s overloaded EC2 instances.

AWS References:

Amazon EC2 Auto Scaling Target Tracking

Amazon SQS ApproximateNumberOfMessages

UsingAD ConnectorwithAWS IAM Identity Center(formerly AWS Single Sign-On) allows the company to leverage its existingon-premises Active Directoryfor centralized identity management and access control. AD Connector acts as a proxy to the on-premises AD withoutrequiring additional infrastructure or complex setup. This solution integrates seamlessly with AWS, allowing the development team to use their existing AD credentials to access AWS resources across multiple accounts managed by AWS Organizations. The permissions for AWS resources can be managed centrally through IAM Identity Center by configuring permission sets.

This solution provides:

Least operational overhead: AD Connector is fully managed, and IAM Identity Center allows centralized management of permissions across accounts.

Secure access: The solution complies with security policies by using existing AD authentication mechanisms.

Option A (AWS Managed AD): Setting up a fully managed AWS AD and establishing a trust is more complex and involves additional operational overhead.

Option B (IAM Users): Manually managing IAM users and permissions is less scalable and increases operational complexity.

Option D (Cognito): Amazon Cognito is more suited for user-facing applications rather than internal identity management for AWS resources.

AWS References:

AD Connector with IAM Identity Center

AWS IAM Identity Center

The most scalable and managed solution for streaming ingestion, real-time transformation, and delivery to Amazon S3 is Amazon Kinesis Data Streams, Amazon Managed Service for Apache Flink, and Amazon Kinesis Data Firehose.

From AWS Documentation:

“Amazon Kinesis Data Streams enables real-time processing of streaming data at massive scale. With Apache Flink on Kinesis Data Analytics, you can process data streams in near real-time, then use Amazon Kinesis Data Firehose to reliably deliver that data to S3.”

(Source: Amazon Kinesis Developer Guide)

Why A is correct:

Fully managed: All services involved are serverless and managed.

Real-time ingestion: Kinesis Data Streams supports sub-second latency and can handle high-throughput workloads like 100,000+ devices.

Near real-time processing: Apache Flink is designed for continuous stream processing with complex event handling.

Efficient delivery: Kinesis Firehose delivers processed data directly to S3 with retry and backup capability.

Why other options are incorrect:

Option B: SQS is not optimized for real-time streaming at high volume.

Option C: EC2 + Kafka + EMR adds high operational overhead and cost.

Option D: EventBridge is event-driven, not designed for high-throughput streaming; AWS Batch is unsuitable for near real-time processing.

To ensure container images are secure and to notify administrators about vulnerabilities, the solution needs (1) a vulnerability scanning capability for container images and (2) a notification mechanism that alerts when findings occur. Amazon Inspector provides automated security assessments and can scan container images stored in Amazon ECR to identify software vulnerabilities and unintended network exposure patterns, producing findings that can be acted upon. Therefore, D addresses the core requirement of detecting vulnerabilities in container images.

To notify administrators with minimal custom work, AWS Config can help by evaluating resources against desired configurations and integrating with notifications through AWS services (for example, via Amazon SNS using Config rules/conformance packs). Using the AWS Config conformance pack for Amazon ECS helps establish a managed set of compliance checks aligned to ECS-related best practices. While Inspector is the system that detects vulnerabilities, Config can be used to enforce and monitor governance controls around the container environment and can trigger notifications when noncompliance is detected. In exam patterns, pairing an Inspector detection capability with a managed governance/notification framework like Config is a common “two-part” answer.

The other options do not meet the requirement: A (privileged mode) can increase risk rather than improve image security; logging does not equal vulnerability detection. C is unrelated because AWS WAF protects web applications at the edge and does not scan container images for CVEs. E is incorrect because Parameter Store stores configuration data and secrets; it does not encrypt container images (ECR encryption at rest is handled by AWS-managed mechanisms and KMS integration, not Parameter Store).

So D provides scanning and B supports managed compliance/notification controls with low operational overhead.

For Amazon RDS relational databases experiencing read-heavy workloads, AWS best practice is to create read replicas and offload read traffic to those replicas. A read replica:

Uses asynchronous replication from the primary.

Can serve read-only queries, thereby reducing load and query latency on the primary.

Can be used behind an application-level read/write splitting mechanism or via endpoints that direct read traffic to replicas.

Performance Insights (Option A) helps diagnose performance problems but does not itself offload traffic.

DAX (Option B) is a cache for DynamoDB, not RDS.

Kinesis Data Firehose (Option D) is a streaming ingestion service and cannot increase RDS query concurrency.

Explanation (AWS Docs):

DynamoDB Accelerator (DAX) is a fully managed, in-memory cache specifically for DynamoDB. It reduces read load and latency without requiring code changes (only SDK config). This is the least development effort.

“Amazon DynamoDB Accelerator (DAX) is a fully managed, highly available in-memory cache for DynamoDB that delivers microsecond read performance and requires minimal application changes.”

— Amazon DAX