A global ad-tech corporation is deploying a real-time bidding (RTB) exchange on AWS. The p… — AWS Certified Solutions Architect - Professional

A global ad-tech corporation is deploying a real-time bidding (RTB) exchange on AWS. The platform must process up to $800,000$ incoming bid requests per second during peak traffic periods, with an end-to-end processing latency budget of $30\text{ ms}$ per request. The architecture requires querying a $12\text{ TB}$ user profile store with sub-millisecond read latency to enrich each bid request. In addition, the system must capture and persist transactional bid logs for downstream offline processing, analytics, and compliance audits without blocking the synchronous request-response flow. The platform must dynamically and instantly scale to accommodate sudden traffic bursts.

Which combination of architectural options will satisfy these performance, scalability, and latency requirements? (Select TWO.)

Deploy a Network Load Balancer (NLB) to distribute incoming traffic to the application tier hosted on Amazon ECS with AWS Fargate using the AWS VPC (awsvpc) network mode.Answer
Use Amazon DynamoDB with Amazon DynamoDB Accelerator (DAX) to store user profile data, configuring the application to query the DAX cluster for user attributes.Answer
C
Deploy an Application Load Balancer (ALB) to distribute incoming traffic, and configure AWS Auto Scaling with Target Tracking policies based on the RequestCountPerTarget metric to scale the backend tasks during sudden traffic spikes.
D
Store the user profile store in an Amazon RDS for PostgreSQL database configured with Multi-AZ, and configure the application to direct read queries to the standby instance in the secondary Availability Zone.
E
Invoke synchronous AWS Lambda functions behind an Amazon API Gateway to evaluate each bid request, utilizing the default regional concurrency pool to handle sudden traffic peaks.

Answer

Deploying a Network Load Balancer (NLB) with Amazon ECS on AWS Fargate using the awsvpc network mode, and storing user profile data in Amazon DynamoDB with DynamoDB Accelerator (DAX).

The correct architecture uses a Network Load Balancer (NLB) combined with Amazon ECS running on AWS Fargate in the AWS VPC (awsvpc) network mode to handle rapid, massive traffic spikes without pre-warming, while providing low-latency container execution. For the data layer, Amazon DynamoDB with Amazon DynamoDB Accelerator (DAX) is the optimal choice for the

12\text{ TB}

user profile store because it provides sub-millisecond read latencies through in-memory caching, fitting within the

30\text{ ms}

end-to-end latency budget.

Step-by-Step Solution

Analyze the load balancing and traffic burst requirements.

Identify that the platform experiences rapid, volatile traffic spikes up to

800,000

requests per second. A Network Load Balancer (NLB) must be selected over an Application Load Balancer (ALB) because NLB can scale instantly to handle millions of requests without requiring pre-warming or incurring scaling latency.

ALB requires pre-warming to handle sudden, large traffic spikes, which fails the requirement of handling unpredictable bursts automatically.

Evaluate the compute tier for performance and scalability.

Select Amazon ECS on AWS Fargate using the awsvpc network mode.

Fargate provides serverless, scalable container execution, while the awsvpc network mode assigns a dedicated ENI to each task, eliminating port address translation overhead and maximizing network throughput.

Evaluate the data store for latency and capacity constraints.

Select Amazon DynamoDB with DynamoDB Accelerator (DAX) to host the

12\text{ TB}

user profile dataset.

DAX provides a fully managed, highly available in-memory cache in front of DynamoDB, reducing read latencies to sub-milliseconds, which is necessary to meet the

30\text{ ms}

end-to-end latency budget.

Key Concept

Selecting high-throughput, low-latency architectures that can scale horizontally and instantly to meet strict latency budgets under unpredictable load.

Estimated Time:3m 0s

Question

Answer

Step-by-Step Solution

Key Concept

Common Mistakes

Topics

More questions from this topic