SQS Standard queues provide at-least-once delivery; they can, by design, deliver duplicate messages, and you cannot turn that off.

To guarantee no duplicates to consumers, AWS provides SQS FIFO queues with exactly-once processing semantics in conjunction with deduplication.

The minimal change to achieve this behavior is to:

Recreate the queue as a FIFO queue and

Enable content-based deduplication, so SQS automatically deduplicates messages with identical bodies within the deduplication window.

Why others are not correct:

A: Long polling reduces empty responses and costs but does not eliminate duplicates.

C: Content-based deduplication is available only for FIFO queues, not for standard queues—so you cannot “modify” a standard queue this way.

D: Moving to Amazon MQ is a much bigger architectural change and adds more operational overhead; it’s not the “fewest changes” approach.

Amazon RDS for MySQL supports read replicas that offload read-intensive workloads such as reporting, leaving the primary instance free for write operations.

“You can use Amazon RDS read replicas to elastically scale out beyond the capacity constraints of a single DB instance for read-heavy database workloads.”

— Working with Read Replicas

This is the recommended approach for minimizing performance impact on the primary DB instance.

Amazon CloudFront is a global content delivery network (CDN) that caches and distributes content to users with low latency. By setting the existing ALB as the origin, CloudFront can cache dynamic and static content closer to users worldwide, improving performance and reducing the load on the application servers. This is the most cost-optimized and AWS-recommended way to globally serve dynamic content for web applications.

Reference Extract:

" CloudFront accelerates delivery of both static and dynamic content, reducing latency and offloading origin resources by caching content at edge locations. "

Source: AWS Certified Solutions Architect – Official Study Guide, CloudFront and Global Application section.

AWS WAF has limits on how much of an HTTP request body it can inspect. When requests exceed the inspectable size, AWS WAF treats the body as oversize relative to the configured inspection limits, which can lead to rules not evaluating the entire body content. If suspicious payloads are embedded beyond the inspected portion of a large POST request (for example, > 8 MB), WAF cannot reliably detect them purely through body inspection rules.

Given this constraint, the most effective way to “ensure” proper protection is to implement an oversize handling strategy using AWS WAF rule logic: block oversized requests by default and then explicitly allow oversized requests only from known legitimate sources. Option A accomplishes this by adding a rule that blocks oversize requests (so attackers cannot bypass inspection by sending very large bodies) and a higher-priority allow rule to permit large requests from trusted hosts (for example, specific known partners, internal CIDRs, or authenticated upstream systems). This design reduces the attack surface and provides deterministic behavior for requests that cannot be fully inspected.

Option B is incorrect because Shield Advanced is for DDoS protection and does not extend WAF’s request-body inspection size. Option C is unrelated: Content-Type can influence application parsing, but it will not overcome WAF body-size inspection limitations. Option D is not a practical fit for AWS WAF inspection because WAF evaluates requests at the edge/service layer; it does not natively “call Lambda to rewrite the request body” before WAF evaluates it. Any preprocessing would require a different architectural pattern (such as handling uploads out-of-band), which is beyond the scope and would add operational complexity.

Therefore, A is the correct approach: implement explicit oversized request handling by blocking by default and allowing only vetted large requests.

The optimal solution for low-latency global caching with disaster recovery and cross-Region replication is to use Amazon ElastiCache Global Datastore for Redis OSS.

A Global Datastore enables fully managed cross-Region replication and supports automatic failover by promoting read replica clusters in another Region.

ElastiCache ensures encryption in-transit and at-rest, meeting compliance and security requirements.

It ' s a fully managed AWS-native feature, reducing operational effort compared to setting up DMS-based or snapshot-based replication manually.

Other options (A, C, D):

Require manual setup and management (e.g., custom DMS pipelines, snapshots).

Do not offer real-time replication or failover without manual intervention.

???? Reference:

ElastiCache Global Datastore for Redis

Amazon EC2 On-Demand Instances: Since the batch jobs cannot be disrupted, On-Demand Instances provide the necessary reliability and availability.

Amazon S3 Standard: Storing data in S3 Standard for the first 30 days ensures quick and frequent access.

S3 Glacier Deep Archive: After 30 days, moving data to S3 Glacier Deep Archive significantly reduces storage costs for data that is rarely accessed.

S3 Lifecycle Configuration: Automating the transition and deletion of objects using lifecycle policies ensures cost optimization and compliance with data retention requirements.

AWS advises using serverless event-driven processing to minimize operational burden and scale automatically with demand. Amazon SQS can be directly configured as an event source for AWS Lambda. With this setup, Lambda polls the queue, invokes the function, and processes messages without requiring the customer to manage infrastructure. Scaling is automatic, based on the volume of messages in the queue. Option A (increasing instance size) still leaves scaling and management challenges. Option B reduces cost when idle but does not address scaling. Option D requires manual triggering and does not meet continuous processing requirements. Migrating the script to Lambda (C) fully eliminates the need for instance management, providing scalability, reliability, and reduced operational overhead.

AWS Lambda is not suitable for long-running jobs that can take up to 20 minutes, as Lambda has a maximum execution duration of 15 minutes.

Amazon ECS with AWS Fargate allows you to run containers without managing EC2 instances, providing a scalable and highly available environment. You can scale Fargate tasks to handle large and parallel video processing jobs. Amazon Aurora is a highly available, managed relational database. S3 Intelligent-Tiering is cost-effective for storing large video files with variable access patterns.

AWS Documentation Extract:

" AWS Fargate lets you run containers without managing servers or clusters, providing a highly available and scalable compute environment. Fargate is suitable for running data processing workloads that require long-running compute tasks. "

(Source: AWS Fargate documentation, Use Cases)

A: Lambda is not suitable for long-running (over 15 min) or heavy compute jobs.

C: Amazon EMR is optimized for big data analytics, not specific video processing. FSx for Lustre and Kinesis are not best fit for this use case.

D: EKS with EC2 adds operational overhead and RDS in a single AZ is not highly available; Glacier Deep Archive is not suitable for frequently accessed video files.

Amazon S3 is the natural long-term storage service for terabytes of images and videos because it scales massively and integrates directly withAmazon SageMaker AIfor machine learning workflows. UsingS3 Intelligent-Tieringhelps optimize storage cost automatically when access patterns vary over time. AWS recommends Intelligent-Tiering when access frequency changes or is unpredictable, because the service automatically moves objects between tiers while keeping them available. The alternatives introduce extra file system copies and additional cost or use a notebook instance as a storage endpoint, which is not a scalable long-term ingestion design. Therefore, S3 with direct SageMaker integration and Intelligent-Tiering is the most scalable and cost-conscious solution. (AWS Documentation)

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

Amazon S3 uses gateway VPC endpoints, which enable private, secure access to S3 without traversing the internet, compatible with IPv6.

Amazon DynamoDB uses interface VPC endpoints (powered by AWS PrivateLink) for private connectivity within the VPC.

Therefore, for secure private communication without public internet, the correct solution is to implement a gateway VPC endpoint for S3 and an interface VPC endpoint for DynamoDB.

Option B is incorrect because S3 does not support interface endpoints; Option C is incorrect because DynamoDB does not support gateway endpoints. Option D reverses the correct endpoint types.

AWSDirect Connectis the best solution for establishing a consistent, low-latency connection from an on-premises data center to AWS without using the public internet. Direct Connect offers dedicated, high-throughput, and low-latency network connections, which are ideal for performance-sensitive applications like a trading platform that processes high volumes of market data and stock trades.

Direct Connect provides a private connection to your AWS VPC, ensuring that data doesn ' t traverse the public internet, which enhances both security and performance consistency.

AWS References:

AWS Direct Connectprovides a dedicated network connection to AWS services with consistent, low-latency performance.

Best Practices for High Performance on AWSfor performance-sensitive workloads like trading platforms.

Why the other options are incorrect:

A. AWS Client VPN: While this offers secure connectivity, it ' s over the public internet and is not designed for the low-latency, high-performance needs of a trading platform.

C. AWS PrivateLink: PrivateLink is used for connecting VPCs and services within AWS, but it is not designed for connecting on-premises data centers to AWS.

D. AWS Site-to-Site VPN: Although this provides secure connectivity, it uses the public internet, which can introduce latency and doesn ' t meet the low-latency requirements of the use case.

Because the application is public and serves global customers with dynamic IP addresses, the web tier must allow HTTPS traffic from0.0.0.0/0on port 443. For the database tier, the secure design is to allow MySQL access only from theweb tier’s security group, not from the internet. AWS RDS documentation explains that you can specify another VPC security group as the source for the DB security group, which allows application servers to connect without exposing the database publicly. That makes A the correct configuration. Options B and C do not work well for global dynamic users, and D exposes the database to the internet.

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

The correct answer is to useIAM roles with fine-grained permissions. AWS recommends temporary credentials and least-privilege access instead of shared accounts or long-term access keys. Assigning a separate role to each developer gives the company strong access isolation, simpler auditing, and the ability to restrict each developer to only the AWS actions they need. Shared IAM user credentials are insecure because actions cannot be tied reliably to one person. Long-term access keys increase credential management risk. Amazon Cognito is primarily for application end users, not for granting internal developers administrative access to AWS resources. Using IAM roles is the standard AWS security best practice for controlled workforce access.

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

Tape Gateway is purpose-built for organizations that want tokeep their existing tape backup workflowswhile moving the tape infrastructure to AWS. AWS documentation states that Tape Gateway provides cloud-backed virtual tape storage and can archive tapes toS3 Glacier Flexible Retrieval or S3 Glacier Deep Archive. Because the company needs 10-year retention and wants the most cost-effective archive tier,Glacier Deep Archiveis the best fit. This approach also avoids redesigning the backup process, which is one of the explicit requirements. Snowball can move data in bulk, but it changes the workflow and the storage-class target in option C is not the lowest-cost long-term archive. Tape Gateway aligns directly with both operational continuity and archival economics.

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

Throughput Optimized HDD (st1) is the correct and most cost-effective EBS choice because the workload reads and processes large log files sequentially. AWS defines st1 volumes as low-cost magnetic storage designed for frequently accessed, throughput-intensive workloads, including log processing. This workload does not require high random IOPS or ultra-low latency, so gp3 and io1 would be more expensive than necessary. Cold HDD (sc1) is cheaper but intended for less frequently accessed data, so it is not appropriate for an active log-processing application. Because the company wants minimal architectural change from SAN-based block storage, an EBS volume preserves the block-storage model while meeting the sequential throughput requirement.

Amazon RDS Proxyis the correct answer because it is specifically designed to improve connection handling for relational databases, including Aurora. AWS documentation explains that RDS Proxy providesconnection pooling, helps handleunpredictable surges in database traffic, and automatically determines the current Aurora writer instance. That reduces the stress of large numbers of short-lived connections and improves application behavior during failover events because clients reconnect through the proxy rather than directly to database endpoints. Switching database engines would be unnecessary and disruptive. DAX is for DynamoDB, not Aurora. Therefore, placingRDS Proxyin front of the Aurora cluster is the most direct way to reduce connection pressure and improve failover behavior. (AWS Documentation)

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

Explanation (AWS Docs):

DynamoDB has a built-in feature called Time to Live (TTL) which automatically deletes expired items without manual intervention. This requires adding a timestamp attribute and setting a TTL on the table. This is the lowest operational overhead approach.

“You can use DynamoDB TTL to automatically delete items after a specified time, reducing storage costs and administrative overhead.”

— DynamoDB TTL

Amazon S3 Standard-IA: This storage class is designed for data that is accessed less frequently but requires rapid access when needed. It offers lower storage costs compared to S3 Standard while still providing high availability and durability.

Access Patterns: Since the files are frequently accessed in the first 30 days and rarely accessed afterward, transitioning them to S3 Standard-IA after 30 days aligns with their access patterns and reduces storage costs significantly.

Lifecycle Policy: Implementing a lifecycle policy to transition the files to S3 Standard-IA ensures automatic management of the data lifecycle, moving files to a lower-cost storage class without manual intervention. Deleting the files after 4 years further optimizes costs by removing data that is no longer needed.

The requirements include:

Scalability: The solution must scale as video files are uploaded.

Long-running tasks: Processing tasks can take up to 30 minutes. AWS Lambda has a maximum execution time of 15 minutes, which rules out options that involve Lambda performing all the processing.

Serverless and event-driven architecture: Ensures cost-effectiveness and high availability.

Analysis of Options:

Option A:

AWS Lambda has a 15-minute timeout, which cannot support tasks that take up to 30 minutes.

EventBridge Scheduler is unnecessary for monitoring files when native event notifications are available.Not a valid choice.

Option B:

AWS Step Functions and AWS Fargate can handle long-running processes, but Amazon EFS is not the ideal storage for uploaded video files in a serverless architecture.

Processing tasks triggered by EFS events are not a common pattern and may introduce complexities.Not the best practice.

Option C:

Amazon S3 is used for storing uploaded files, which integrates natively with event-driven services like EventBridge and Step Functions.

Amazon S3 event notifications trigger a Step Functions workflow, which can orchestrate Fargate tasks to process large video files, meeting the scalability and execution time requirements.Correct choice.

Option D:

Similar to Option A, AWS Lambda cannot handle long-running processes due to its 15-minute timeout.

Invoking Lambda for processing directly is not feasible for tasks that take up to 30 minutes.Not a valid choice.

AWS References:

Amazon S3 Event Notifications:AWS Documentation - S3 Event Notifications

AWS Step Functions:AWS Documentation - Step Functions

AWS Fargate:AWS Documentation - Fargate

Comparison of Storage Services:AWS Storage Options

By leveragingAmazon S3,Step Functions, andFargate, this solution provides a scalable, efficient, and serverless approach to handling video processing tasks.

Cluster Placement Group: This type of placement group is designed to provide low-latency network performance and high throughput by grouping instances within a single Availability Zone. It is ideal for applications that require tightly coupled node-to-node communication.

Configuration:

When launching EC2 instances, specify the option to launch them in a cluster placement group.

This ensures that the instances are physically located close to each other, reducing latency and increasing network throughput.

Benefits:

Low-Latency Communication: Instances in a cluster placement group benefit from enhanced networking capabilities, enabling low-latency communication.

High Network Throughput: The network performance within a cluster placement group is optimized for high throughput, which is essential for HPC workloads.