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.