AWS Exam Review

High-Performing Architecture Trap Drills

Practice SAA-C03 performance traps around storage choice, compute elasticity, database access patterns, caching, edge services, networking, ingestion, analytics, and scaling.

intermediate5 min readUpdated 2026-06-05CloudCertificationCapacityDataNetworkingTradeoffs
Performance BottleneckStorage SemanticsCompute ElasticityDatabase Access PatternCachingEdge AccelerationData IngestionAnalytics Query

After this, you will understand

These drills teach performance as bottleneck classification instead of reflexively picking the largest instance or fastest-looking service.

Plain version

Each drill gives a workload symptom and asks which AWS service or design matches the access pattern.

Decision pressure

Learners confuse object, block, and file storage, choose Redshift for ad hoc S3 log queries, or use a cache without knowing the freshness tradeoff.

Exam-ready model

Name the constrained layer, identify the access pattern, then choose the service, scaling model, or cache that matches it.

Think before readingWhat is the most common performance-drill mistake?
Scaling the visible compute layer while the real bottleneck is storage semantics, database access pattern, network path, or data layout.

Reading in progress

This page is saved in your local study history so you can continue later.

Next: Design Cost-Optimized Architectures

Study path

Read these in order

Start with the mechanics, then move into the patterns that explain why the system is shaped this way.

  1. 1Cost-Optimized Architecture Trap DrillsAWS Review
  2. 2Athena vs Redshift vs OpenSearchaws-services

Concepts Covered

  • Performance practice drills
  • Storage service selection
  • Compute elasticity
  • Database access patterns
  • Caching and read scaling
  • Edge acceleration
  • Network performance
  • Streaming and queues
  • Analytics service selection
  • SAA-C03 distractor patterns

1. Domain Mental Model

Performance traps are usually wrong-bottleneck traps.

The drill habit is:

find the pressure, then match the service boundary

Do not start with "fastest service." Start with the workload shape. Does it need object storage, block storage, shared files, low-latency key-value access, relational joins, full-text search, streaming ingestion, event routing, caching, CDN delivery, or static global IP entry?

AWS has many high-performing services because workloads have different bottlenecks.

2. Official Task Map

This drill page maps to the high-performing architecture domain:

  • high-performing and scalable storage
  • high-performing and elastic compute
  • high-performing database solutions
  • scalable network architectures
  • data ingestion and transformation

The exam often disguises these as business statements: lower latency, handle spikes, process logs, support global users, query data in S3, scale reads, reduce database load, or choose the best storage type.

Each statement points to a different performance layer.

3. What AWS Is Testing

AWS is testing whether you know service fit and scaling behavior.

S3 scales object storage, but it is not a block device. EBS is block storage for EC2, but it is not a shared file system for many instances. EFS is shared NFS, but it is not the default database volume. DynamoDB scales key-value access, but it is not a relational join engine. Athena queries files in S3, but it is not low-latency full-text search. CloudFront caches content, but it does not provide fixed global IPs.

These drills train service-boundary clarity.

4. Service And Concept Clusters

Use this cluster map while drilling:

5. Architecture Reasoning Patterns

Use this drill checklist:

1. Is the bottleneck read, write, compute, network, ingestion, query, or storage semantics?
2. Is latency, throughput, concurrency, or data volume the dominant issue?
3. Is the access pattern predictable?
4. Can stale reads be tolerated?
5. Is the data structured for the chosen query engine?
6. Does the solution scale horizontally, vertically, or through managed capacity?

For cache questions, ask what can be stale and who invalidates it.

For analytics questions, ask where the data already lives and what query shape is needed.

For global traffic questions, ask whether the issue is cacheable content, dynamic traffic routing, fixed IPs, or DNS policy.

6. High-Yield Comparisons

Drill 1: many EC2 instances need shared Linux files.

Wrong instinct: EBS.

Better answer: EFS.

Drill 2: user-uploaded images need durable storage.

Wrong instinct: instance store or EBS on web servers.

Better answer: S3.

Drill 3: ad hoc SQL over CloudTrail logs in S3.

Wrong instinct: Redshift cluster.

Better answer: Athena with Glue catalog and partitioning.

Drill 4: full-text application search.

Wrong instinct: Athena.

Better answer: OpenSearch.

Drill 5: static web assets for global users.

Wrong instinct: Global Accelerator.

Better answer: CloudFront.

Drill 6: fixed global IPs for TCP/UDP app.

Wrong instinct: CloudFront.

Better answer: Global Accelerator.

7. Scenario Triggers

"NFS" points to EFS.

"Block storage attached to EC2" points to EBS.

"Temporary high-speed local scratch" points to instance store.

"Unknown object access pattern" may point to S3 Intelligent-Tiering.

"Scale key-value reads and writes" points to DynamoDB.

"Relational transactions and SQL" points to RDS or Aurora.

"Reduce repeated database reads" points to ElastiCache or DAX.

"Stream records continuously" points to Kinesis.

"Buffer work between producer and consumer" points to SQS.

"Route events between applications" points to EventBridge.

8. Common Traps

Do not choose EBS for general multi-instance shared files.

Do not choose EFS for object uploads better served by S3.

Do not choose S3 when the OS needs a block device.

Do not use read replicas for write scaling.

Do not use cache as a source of truth.

Do not choose Redshift for occasional S3 file queries.

Do not choose OpenSearch for warehouse-style BI joins.

Do not choose CloudFront because the word global appears.

Do not ignore partitioning, compression, and file size in Athena performance.

Do not scale compute before checking database and network bottlenecks.

9. Study Path

Study and drill in this order:

  1. Design High-Performing Architectures
  2. S3 vs EBS vs EFS vs Instance Store
  3. Lambda vs ECS/Fargate vs EC2
  4. DynamoDB vs RDS vs Aurora
  5. ElastiCache vs DynamoDB DAX
  6. CloudFront vs Global Accelerator
  7. Kinesis vs SQS vs EventBridge
  8. Athena vs Redshift vs OpenSearch
  9. Analytics Data Lake On S3
  10. Serverless API With Lambda And DynamoDB

Repeat until scenario wording immediately reveals the bottleneck layer.

Review Design High-Performing Architectures, Secure Architecture Trap Drills, Resilient Architecture Trap Drills, and Cost-Optimized Architecture Trap Drills.

Official AWS references:

What to study next

These links keep the session moving: read prerequisites first, then open the systems, concepts, and patterns that deepen this page.