Examkin

Question

Difficulty: MediumCost-Optimized Compute Selection and Purchasing Strategies

A game development company is designing the infrastructure for a new multiplayer mobile game on AWS. The application has the following requirements:

* A game API service that experiences highly variable, spiky traffic throughout the day.
* A matchmaking service that must run continuously 2424 hours a day, 77 days a week on Amazon EC2.
* A nightly batch processing job that processes game telemetry, is non-time-critical, can tolerate interruptions, and produces temporary output files that are deleted within 4848 hours.
* A relational database tier running on Amazon RDS for MySQL.

Which combination of hosting configurations and purchasing strategies is the most cost-effective for this workload?

  1. Host the game API on AWS Lambda, purchase a Compute Savings Plan for the matchmaking EC2 instances, run the batch processing job on EC2 Spot Instances storing temporary files in Amazon S3 Standard, and purchase Reserved Instances for the Amazon RDS database.Answer
  2. B
    Host the game API on AWS Lambda, purchase a single Compute Savings Plan to cover both the matchmaking EC2 instances and the Amazon RDS database, and run the batch processing job on EC2 Spot Instances storing temporary files in Amazon S3 Standard.
  3. C
    Host both the game API and the matchmaking service on AWS Lambda, run the batch processing job on EC2 Spot Instances storing temporary files in Amazon S3 Standard, and purchase Reserved Instances for the Amazon RDS database.
  4. D
    Host the game API on AWS Lambda, purchase a Compute Savings Plan for the matchmaking EC2 instances, run the batch processing job on EC2 Spot Instances storing temporary files in Amazon S3 Standard-Infrequent Access (S3 Standard-IA), and purchase Reserved Instances for the Amazon RDS database.

Answer

Host the game API on AWS Lambda, purchase a Compute Savings Plan for the matchmaking EC2 instances, run the batch processing job on EC2 Spot Instances storing temporary files in Amazon S3 Standard, and purchase Reserved Instances for the Amazon RDS database.
Hosting the spiky game API on AWS Lambda ensures payment is only for actual requests. The continuously running matchmaking EC2 instances are discounted using a Compute Savings Plan. The interruptible nightly batch job runs on low-cost Spot Instances, and storing its short-lived files in Amazon S3 Standard avoids the 3030-day minimum billing penalty of S3 Standard-IA. Finally, purchasing Reserved Instances for the Amazon RDS database reduces its continuous running costs.

Step-by-Step Solution

1
Analyze the workload characteristics of each application tier.
Identify the game API as spiky/variable, the matchmaking service as continuous 2424 hours a day, 77 days a week, the batch job as interruptible and short-lived (4848 hours), and the database as continuous relational (RDS).
Matching workload patterns to the correct compute and storage options is the foundation of cost optimization.
2
Select the most cost-effective hosting and purchasing model for each tier.
Assign AWS Lambda to the spiky API, EC2 with a Compute Savings Plan to the continuous matchmaking service, EC2 Spot Instances with Amazon S3 Standard to the batch job, and RDS Reserved Instances to the database.
Compute Savings Plans cover EC2 but not RDS. Spot Instances are perfect for fault-tolerant batch workloads. S3 Standard is cheaper for short-term files (less than 3030 days) than S3 Standard-IA due to minimum storage duration charges.

Key Concept

Workload-based compute selection, Savings Plans scope, Spot Instances, and S3 storage tier cost structures.
Rate this question

Topics

Cost-Optimized Compute Selection and Purchasing Strategies