ExamworthyExamworthy
Open the appStart free

SAA-C03 - Design High-Performing Architectures - Section 3.3

Design high-performing and elastic compute by selecting EC2 instance types, AWS Lambda and container compute.

Describe the EC2 instance families - general purpose, compute optimised, memory optimised, storage optimised, and accelerated computing - and match each to workload characteristics such as high CPU-to-memory ratio or GPU requirements. Distinguish AWS Fargate serverless container execution from self-managed EC2-backed ECS clusters, and recognise when AWS Lambda eliminates instance management overhead for short-duration event-driven tasks.

EC2 instance familiesAWS FargateAWS LambdaCompute right-sizing

Practice question for this objective

Free sampleDesign High-Performing Architecturesmedium

A startup is modernising a monolith into two independent workloads. The first is a stateless image-resizing task that fires thousands of times in short bursts when objects land in Amazon S3, each run finishing in well under a second, with long idle gaps in between. The second is a containerised nightly reconciliation job that runs for about 40 minutes, is triggered once a day, and must not be interrupted partway through. The team wants the least operational overhead for each workload, refuses to provision or patch any EC2 host fleet, and wants to pay only for resources actually consumed. Which TWO compute choices best fit these workloads? Select TWO.

  • ARun the image-resizing task on AWS Lambda, invoked by Amazon S3 event notifications, so it scales out per event and bills only for execution time. Correct
  • BRun the image-resizing task on a fixed fleet of On-Demand EC2 instances behind an Auto Scaling group sized for the peak burst.
  • CRun the nightly reconciliation job on AWS Lambda triggered by an Amazon EventBridge schedule, with the function timeout raised to its highest value.
  • DRun the nightly reconciliation container on an Amazon EC2 instance in an Auto Scaling group using Spot Instances to cut cost.
  • ERun the nightly reconciliation container as an Amazon ECS task on AWS Fargate, billed for the vCPU and memory it requests for the run. Correct
Match elastic compute to workload shape: Lambda for short event-driven bursts and Fargate for longer containerised runs that exceed Lambda's timeout. Lambda fits short, spiky, stateless event work because it scales per event and bills only for execution, but its 15-minute timeout rules out the 40-minute job. Fargate runs that longer container with per-task billing and no host fleet, while Spot and standing EC2 fleets reintroduce interruption risk or host management the team rejected.

Why A is correct: Lambda scales out instantly per S3 event, costs nothing while idle, and suits sub-second stateless bursts, so it fits the resizing workload precisely.

Why B is wrong: A standing EC2 fleet still needs patching and pays for idle capacity between bursts, which contradicts the no-host-management and pay-per-use requirements.

Why C is wrong: Lambda is serverless and schedulable, but its 15-minute maximum timeout cannot accommodate a 40-minute run, so the job would be terminated before completing.

Why D is wrong: Spot capacity can be reclaimed with two minutes notice, which would interrupt the uninterruptible job, and the team still has to manage and patch the EC2 host.

Why E is correct: Fargate runs the long container with no host fleet to manage, has no 15-minute cap, and bills per task run, matching the 40-minute uninterruptible job.

See more SAA-C03 practice questions, answers explained.

More in this domain

Back to all Design High-Performing Architectures objectives, or the SAA-C03 cert hub.

Examworthy is not affiliated with or endorsed by Amazon Web Services. Original, blueprint-aligned practice material only.