Examkin

Question

Difficulty: Very hardAWS Compute Services

A cloud architect is reviewing the compute requirements for a media distribution platform. The platform has two distinct components:

1. A video rendering pipeline that requires deep customization of the underlying Linux kernel and direct access to physical GPU hardware resources.
2. A containerized metadata extraction service that runs on an event-driven basis in response to user uploads, with highly unpredictable traffic and a strict requirement to avoid managing host servers.

Which two of the following compute solutions should the architect select to meet these requirements?

  1. Amazon EC2 instances configured with custom Amazon Machine Images (AMIs) to support the required custom operating system kernel and GPU accessAnswer
  2. B
    AWS Lambda to execute the video rendering workload by deploying the custom kernel within a container image
  3. AWS Fargate to run the containerized metadata extraction service dynamically without provisioning or managing host serversAnswer
  4. D
    Amazon EC2 instances where AWS automatically performs all guest operating system patching and kernel configuration updates

Answer

The architect should select Amazon EC2 instances configured with custom Amazon Machine Images (AMIs) to support the custom operating system and GPU requirements, and AWS Fargate to run the containerized metadata extraction service dynamically without provisioning or managing host servers.
Selecting Amazon EC2 with custom AMIs and AWS Fargate is correct because EC2 provides the administrative access needed to customize the operating system kernel and utilize physical GPU instances, while AWS Fargate provides a serverless compute engine to run containerized workloads dynamically without host management.

Step-by-Step Solution

1
Analyze the video rendering pipeline requirements: deep kernel customization and direct GPU hardware access.
Identify that a service granting full control over the guest operating system (IaaS) and supporting GPU hardware instances is required. Amazon EC2 meets these criteria.
Serverless or fully managed services do not allow direct kernel configuration or physical hardware mappings.
2
Analyze the metadata extraction service requirements: containerized, event-driven, unpredictable usage, and no host server management.
Identify that a serverless container orchestration platform is required. AWS Fargate runs containers without requiring instance management.
AWS Fargate manages the underlying compute infrastructure automatically, scaling to meet demand without operational overhead.
3
Evaluate the remaining options against the AWS Shared Responsibility Model and service limitations.
Eliminate the option suggesting AWS patches EC2 guest operating systems, and eliminate AWS Lambda due to its inability to run custom kernels or access physical GPU hardware.
Operating system patching on EC2 is a customer responsibility, and Lambda does not support the rendering workload's hardware and kernel needs.

Key Concept

Selecting appropriate AWS compute services based on operational control, infrastructure management boundaries, and application architecture.
Rate this question

Topics

AWS Compute Services

More questions from this topic