AWS DEA-C01 Practice Test

DEA-C01 practice test: what's covered

• Data Engineer Associate

Free DEA-C01 sample questions

1. A fleet of 800,000 IoT sensors emits telemetry that must be processed in strict per-device order while sustaining roughly 1 million records per second. Which ingestion design preserves ordering and scales to that throughput?

   • A. Send every reading to a single standard SQS queue and rely on the queue's FIFO redelivery to maintain device order
   • B. Use Amazon Kinesis Data Streams with the partition key set to deviceId so records for a device map to one shard
   • C. Push records directly into an Amazon S3 prefix per device and trigger Lambda on each PutObject event
   • D. Buffer records in Amazon DynamoDB and scan the table every minute to reorder them before processing
   Show answer

   Correct: B. Kinesis Data Streams guarantees ordering within a shard, and using deviceId as the partition key gives per-device affinity while shards scale to the required throughput. A standard SQS queue is explicitly not ordered, and SQS FIFO caps at far lower throughput, so it is incorrect. Writing per-device S3 prefixes does not preserve real-time ordering, and a DynamoDB scan-and-reorder pattern is not designed for millions of records per second.

2. Your team must continuously land newly arriving files into a curated S3 zone and automatically kick off a transformation job as each file lands. Which combination of AWS services builds this event-driven pipeline? (Choose TWO)

   • A. Configure S3 Event Notifications with a prefix and suffix filter to invoke a Lambda function or enqueue to SQS
   • B. Enable AWS Glue to run a crawler on a fixed five-minute cron and assume files arrived in that window
   • C. Use Amazon EventBridge rules on S3 object-created events to start an AWS Glue or Step Functions workflow
   • D. Mount the S3 bucket as a local filesystem and watch it with an inotify daemon on an EC2 instance
   • E. Poll the bucket every second with a custom script that compares object listings to detect new keys
   Show answer

   Correct: A and C. S3 Event Notifications and EventBridge object-created events both deliver per-object triggers, so they correctly drive transformation jobs the moment a file lands. A fixed cron crawler is not event-driven and will miss or duplicate work, so it is incorrect. Mounting S3 with inotify is unreliable because S3 is not a POSIX filesystem, and second-by-second polling wastes LIST requests and adds latency unlike native event delivery.

3. A data analyst with no Spark experience needs to profile a messy CSV dataset, remove duplicate rows, standardize date formats, and publish the cleaned output to S3 without writing code. Which AWS service best fits this no-code preparation requirement?

   • A. AWS Glue Spark ETL jobs authored directly in PySpark scripts
   • B. Amazon EMR clusters running a hand-written Scala Spark application
   • C. AWS Glue DataBrew with its visual recipe and profiling transforms
   • D. Amazon Athena federated queries against the raw CSV in place
   Show answer

   Correct: C. Glue DataBrew is purpose-built for analysts to profile, clean, and normalize data through a visual recipe of over 250 transforms with no code, which matches the requirement exactly. Glue Spark ETL and a Scala EMR application both require writing code, so they are incorrect for a no-code user. Athena federated queries read data but do not perform interactive cleaning and deduplication, unlike DataBrew.

4. An ingestion pipeline writes to a table whose source occasionally adds new columns. The team wants additive schema changes, ACID transactions, and time-travel queries on data stored in S3 and queried by Athena. Which table format should they adopt?

   • A. Plain CSV files registered as an external table in the Glue Data Catalog
   • B. Apache Iceberg tables managed through the Glue Data Catalog and queried by Athena
   • C. Uncompressed JSON objects partitioned only by the ingestion date prefix
   • D. A single wide Parquet file overwritten in full on every pipeline run
   Show answer

   Correct: B. Apache Iceberg natively supports additive schema evolution, ACID transactions, and snapshot-based time travel, and it is first-class in Athena and Glue, making it the correct choice. Plain CSV and uncompressed JSON have no transaction or time-travel semantics, so they are incorrect. Overwriting a single Parquet file in full destroys history and provides no ACID guarantees, unlike Iceberg's snapshot model.

5. A company runs an Amazon Aurora MySQL OLTP database and wants near-real-time replication of changes into Amazon Redshift for analytics, with minimal pipeline code to maintain. Which approach meets the requirement most efficiently?

   • A. Schedule a nightly full Aurora snapshot and restore it into a fresh Redshift cluster each morning
   • B. Use the Aurora zero-ETL integration to replicate changes into Redshift continuously without custom ETL
   • C. Export Aurora to CSV on S3 every hour and COPY the files into Redshift on a cron schedule
   • D. Enable Redshift query federation and join live against Aurora tables for every analytical query
   Show answer

   Correct: B. Aurora zero-ETL integration replicates committed changes into Redshift with seconds of lag and no ETL pipeline to build or operate, which is the most efficient fit. Nightly snapshot restores and hourly CSV exports introduce hours of staleness and operational overhead, so they are incorrect. Federated queries hit the OLTP database for every analytical query, adding load and latency, unlike the dedicated zero-ETL replica.

Last updated: 2026-10-19

DEA-C01 practice test FAQ

Related practice tests

• AWS SAA-C03 practice test
• AWS MLA-C01 practice test
• Microsoft DP-700 practice test
• Databricks Data Engineer Associate practice test
• SnowPro Core COF-C02 practice test
• Google Cloud Professional Data Engineer practice test