A company runs a high-performance simulation platform on AWS. The application architecture… — AWS Certified Solutions Architect - Associate

A company runs a high-performance simulation platform on AWS. The application architecture has three primary workloads:

1. A front-end web portal and API gateway running on Amazon ECS with AWS Fargate. This has a constant, predictable baseline demand of $15\text{ vCPUs}$ and $30\text{ GB}$ of memory $24/7$ , with brief, unpredictable spikes up to $60\text{ vCPUs}$ during peak daytime hours.
2. A cluster of backend worker nodes running simulation jobs on Amazon EC2. These jobs are stateless, run for $4$ to $6$ hours, can be interrupted and resumed at any time, and are triggered dynamically based on queue depth.
3. An Amazon RDS for MySQL database that stores application state and runs continuously.

Which two strategies should a solutions architect recommend to optimize the compute and database costs for this platform? (Select TWO.)

Purchase a Compute Savings Plan to cover the baseline usage of the ECS Fargate tasks.Answer
Configure an EC2 Auto Scaling group using Spot Instances for the simulation worker nodes.Answer
C
Purchase a Compute Savings Plan to cover the compute costs of the ECS Fargate tasks and the Amazon RDS database instance.
D
Migrate the simulation worker jobs to AWS Lambda to benefit from pay-per-use sub-second billing.
E
Migrate the relational database to Amazon DynamoDB and configure Provisioned Capacity Mode to manage highly unpredictable, spiky traffic.

Answer

To optimize the costs of this architecture, the company should purchase a Compute Savings Plan to cover the baseline usage of the ECS Fargate tasks and configure an EC2 Auto Scaling group using Spot Instances for the simulation worker nodes.

Purchasing a Compute Savings Plan covers the constant, predictable baseline usage of the ECS Fargate tasks at a significantly discounted rate. Using Spot Instances for the simulation worker nodes is highly cost-effective because the workloads are stateless, queue-driven, and can tolerate interruptions, allowing the company to leverage up to

90\%

discounts compared to On-Demand pricing.

Step-by-Step Solution

Analyze the workload characteristics of each compute tier to identify baseline vs. variable usage.

The ECS Fargate web tier has a continuous, predictable baseline of

15\text{ vCPUs}

and

30\text{ GB}

of memory

24/7

. The simulation worker nodes are stateless, run for

4

6

hours, and are queue-driven, meaning they can tolerate interruption.

Understanding the compute characteristics and resilience of each tier is necessary to match them to the correct AWS pricing models.

Select the optimal purchasing model for the continuous, predictable baseline of the ECS Fargate tasks.

Select a Compute Savings Plan, which applies to ECS Fargate, AWS Lambda, and Amazon EC2 usage. This commitment-based model provides substantial discounts over On-Demand rates for the steady-state baseline.

Compute Savings Plans are flexible, apply automatically to Fargate across any region, and lower the cost of continuous container workloads.

Select the optimal purchasing model for the stateless, queue-driven backend simulation worker nodes.

Configure an EC2 Auto Scaling group using Spot Instances to run the backend worker nodes.

Since the simulation jobs are stateless and run from a queue, they are highly tolerant of EC2 instance reclamation. Spot Instances offer the lowest cost (up to

90\%

savings over On-Demand) for interruptible workloads.

Key Concept

Selecting the most cost-effective compute purchasing strategies based on workload predictability, duration, and fault tolerance.

Question

Answer

Step-by-Step Solution

Key Concept

Common Mistakes

Topics