A financial technology firm runs a high-frequency risk assessment platform on AWS. The pla… — AWS Certified Solutions Architect - Associate

A financial technology firm runs a high-frequency risk assessment platform on AWS. The platform consists of the following components:

* A web portal and API gateway hosted on AWS Fargate that requires a continuous baseline capacity of $25\text{ vCPUs}$ and $50\text{ GB}$ of RAM $24/7$ to handle dashboard queries.
* A backend distributed processing grid of Amazon EC2 instances that runs short-lived, containerized simulation jobs. These jobs are highly parallel, run for $2$ to $3$ hours at a time, write checkpoints to an external storage layer, and must complete within a flexible $18$ -hour daily window.
* A database layer hosted on Amazon DynamoDB that stores user session tokens and transaction metadata. The database experiences sudden, unpredictable traffic spikes up to $10,000\text{ requests per second}$ , followed by several hours of idle state.

The firm wants to optimize its architecture to achieve the maximum possible cost savings for its compute resources. Which two options should the solutions architect recommend to satisfy these requirements at the lowest cost?

Purchase a Compute Savings Plan to cover the baseline usage of the AWS Fargate tasks.Cevap
Provision the distributed grid of Amazon EC2 instances using Spot Instances.Cevap
C
Purchase a Compute Savings Plan to cover the baseline usage of both the AWS Fargate tasks and the Amazon DynamoDB table capacity.
D
Refactor the API gateway and web portal backend to run on AWS Lambda to eliminate idle compute costs.
E
Configure the Amazon DynamoDB table in Provisioned Capacity Mode with static capacity scaled to the peak of $10,000\text{ requests per second}$ .

Cevap

The correct recommendations are to purchase a Compute Savings Plan to cover the baseline AWS Fargate API gateway usage, and to provision the EC2 distributed simulation grid using Spot Instances.

The correct configuration is to purchase a Compute Savings Plan to cover the steady-state, continuous AWS Fargate tasks (since Fargate is covered by Compute Savings Plans) and to use Spot Instances for the short-lived, checkpointed, and flexible EC2 simulation jobs (which can handle interruptions without losing progress).

Adım Adım Çözüm

Analyze the web portal and API gateway compute requirements.

The API gateway has a continuous baseline capacity of

25\text{ vCPUs}

and

50\text{ GB}

of RAM running

24/7

. This constant workload is best optimized using a Compute Savings Plan, which applies to Fargate and offers significant discounts over On-Demand pricing.

Baseline, predictable compute workloads running continuously should be covered by a purchasing commitment tool like Savings Plans.

Analyze the backend simulation processing grid compute requirements.

The simulation jobs are short-lived (

2

3

hours), stateless/checkpointed, and can tolerate interruptions within an

18

-hour daily processing window. This profile is ideal for Spot Instances, which provide up to

90\%

savings compared to On-Demand pricing.

Fault-tolerant, stateless, and time-flexible workloads should be run on Spot Instances to minimize compute costs.

Analyze the database layer performance and load characteristics.

The DynamoDB database has highly spiky, unpredictable workloads (

10,000\text{ requests per second}

peak down to

0

for hours). Using On-Demand capacity mode avoids over-provisioning during idle hours, while Compute Savings Plans cannot be applied to DynamoDB.

Compute Savings Plans only apply to EC2, Fargate, and Lambda, not database services like DynamoDB, RDS, or Redshift.

Anahtar Kavram

Selecting the optimal AWS compute purchasing models (Compute Savings Plans, Spot Instances, On-Demand) and database capacity modes to minimize overall cloud architecture cost.

Soru

Cevap

Adım Adım Çözüm

Anahtar Kavram

Sık Yapılan Hatalar

Konular