☁︎SAA-C03

Week3 Storage Db — SAA-C03 Mock Test

Mock Test — Week 3 Storage & Databases (15 questions)

  • Date: 2026-06-08
  • Topics: S3, Storage Gateway, Snow Family, RDS, Aurora, DynamoDB, ElastiCache, Redshift, Athena, Glue, EMR, Neptune, DocumentDB, Keyspaces, Timestream
  • Difficulty: exam-realistic
  • Suggested time: ~25 min

Q1.

A company must retain backup objects in S3 for 7 years of compliance. Restores are extremely rare (once a year, and 48-hour recovery is acceptable). Which storage class is most cost-effective?

  • A) S3 Standard-IA
  • B) S3 Glacier Instant Retrieval
  • C) S3 Glacier Flexible Retrieval
  • D) S3 Glacier Deep Archive

Q2.

A solutions architect must ensure that all new objects in an S3 bucket are encrypted with a customer-managed KMS key. Which combination achieves this?

  • A) Enable default encryption with SSE-S3 and an S3 inventory job
  • B) Enable default encryption with SSE-KMS and a bucket policy denying uploads without the correct KMS key header
  • C) Use S3 Object Lock with Compliance mode
  • D) Enable Block Public Access

Q3.

A web app caches DB query results to serve sub-millisecond reads and reduce database load. Which service fits best?

  • A) DynamoDB DAX
  • B) Amazon ElastiCache for Redis
  • C) Amazon RDS read replica
  • D) Amazon OpenSearch Service

Q4.

A globally distributed application requires multi-region, multi-active writes with single-digit-millisecond latency. Which AWS service is purpose-built for this?

  • A) Aurora Global Database (writer in one region only)
  • B) DynamoDB Global Tables
  • C) RDS for PostgreSQL with cross-region read replicas
  • D) DocumentDB cross-region replicas

Q5.

An e-commerce site's relational DB primary must fail over within a minute if the AZ fails, and the standby need not be readable. Which configuration?

  • A) RDS Single-AZ with frequent snapshots
  • B) RDS Multi-AZ deployment (classic)
  • C) Two RDS instances in different AZs with custom failover scripts
  • D) DynamoDB

Q6.

A Lambda function bursts to 5,000 concurrent executions, overwhelming an Aurora DB with connections. What is the best AWS-native mitigation?

  • A) Add Aurora read replicas
  • B) Migrate to DynamoDB
  • C) Put RDS Proxy in front of the Aurora cluster
  • D) Increase Aurora instance size

Q7.

A company has a 200 TB on-premises data set to migrate to S3. The internet link is 50 Mbps. What is the fastest and most reliable transfer method?

  • A) AWS Direct Connect + DataSync
  • B) Multipart upload over HTTPS
  • C) AWS Snowball Edge devices
  • D) S3 Transfer Acceleration

Q8.

A windows application on-prem uses an SMB share. Management wants the data to live in S3 with continued SMB protocol on-prem and minimal app change. Which service?

  • A) AWS DataSync
  • B) Amazon S3 File Gateway
  • C) Amazon FSx for Windows File Server
  • D) AWS Snowball

Q9.

You need to query 5 TB of historical CloudFront access logs in S3 ad-hoc with SQL, without managing any infrastructure. Which service is best?

  • A) Amazon Redshift cluster
  • B) Amazon Athena
  • C) Amazon EMR Spark
  • D) Aurora Federated Query

Q10.

A team needs to store graph relationships for a recommendation engine (millions of users, products, friendships). Which AWS service is purpose-built?

  • A) Amazon Neptune
  • B) Amazon DocumentDB
  • C) Amazon DynamoDB
  • D) Amazon Keyspaces

Q11.

A DynamoDB table must allow microsecond-level read latency for hot items without changing application code significantly. Which solution fits?

  • A) Front the table with DAX
  • B) Add a Global Secondary Index
  • C) Migrate to ElastiCache Memcached
  • D) Move to Redshift Spectrum

Q12.

A solution architect must protect S3 objects from being accidentally or maliciously deleted for 5 years (write-once-read-many). What should they enable?

  • A) S3 versioning + lifecycle expiration
  • B) Block Public Access
  • C) S3 Object Lock with Compliance mode and a 5-year retention
  • D) Cross-Region Replication

Q13.

A company runs an OLAP workload on 80 TB of data and needs interactive dashboards across multiple concurrent users. Which service fits best?

  • A) RDS Postgres with read replicas
  • B) Amazon Redshift
  • C) Amazon DynamoDB with GSIs
  • D) Amazon Aurora MySQL

Q14.

An IoT pipeline writes 1 million sensor data points per minute. Queries are by device_id + time range. Which database is most cost- and feature-appropriate?

  • A) Amazon Aurora MySQL
  • B) Amazon Redshift
  • C) Amazon Timestream
  • D) Amazon Neptune

Q15.

A company exposes a read-heavy application backed by an RDS MySQL DB. They want to scale read throughput without rewriting the app. Which option is simplest?

  • A) Switch the workload to DynamoDB
  • B) Add RDS read replicas and direct read queries to the reader endpoint
  • C) Enable RDS Multi-AZ
  • D) Increase the instance class to the largest available

Your answers

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.

Answer Key & Explanations — don't peek!

Q1 — Correct: D

Glacier Deep Archive is the cheapest, with 12-48 h retrieval — acceptable for the requirement.

Q2 — Correct: B

Default SSE-KMS encryption plus a bucket policy that denies uploads missing the x-amz-server-side-encryption-aws-kms-key-id header enforces a specific KMS key.

Q3 — Correct: B

ElastiCache is the general-purpose sub-ms cache. DAX (A) is only for DynamoDB workloads.

Q4 — Correct: B

DynamoDB Global Tables is the only AWS service offering multi-region, multi-active writes natively. Aurora Global DB has only one writer.

Q5 — Correct: B

Classic Multi-AZ provides synchronous standby with automatic failover (~60-120 s). Standby not readable, which matches the requirement.

Q6 — Correct: C

RDS Proxy pools connections and accommodates Lambda spikes, while preserving database failover behavior.

Q7 — Correct: C

At 50 Mbps, 200 TB would take many months over the wire. Snowball Edge is the canonical solution for large offline transfers.

Q8 — Correct: B

S3 File Gateway preserves NFS/SMB on-prem while persisting data into S3.

Q9 — Correct: B

Athena = serverless SQL on S3, ad-hoc, pay-per-query — the textbook fit.

Q10 — Correct: A

Neptune is AWS's purpose-built graph database (Gremlin / SPARQL).

Q11 — Correct: A

DAX is the in-memory cache specifically for DynamoDB; it requires minimal client changes via the DAX SDK.

Q12 — Correct: C

Object Lock with Compliance mode + retention period implements WORM; not even the root user can shorten the retention.

Q13 — Correct: B

Redshift is purpose-built for petabyte-scale OLAP with concurrent BI users.

Q14 — Correct: C

Timestream is the managed time-series database, optimized for IoT/telemetry queries.

Q15 — Correct: B

Read replicas scale reads with no app rewrite (just direct reads to replica endpoint). Multi-AZ standby is not readable.