Question bank
AWS Certified Data Engineer – Associate
42 AWS Certified Data Engineer – Associate questions with the correct answer and a written explanation, organised across 4 exam domains.
- 42 questions
- 4 domains
- Foundation level
certilumQuestion index
Jump straight to any question. Answers and explanations are inline.
Data Governance and Security
9 questions in this domain
- Question 1Single Choice
What happens when an IAM policy allows s3:GetObject but the S3 bucket policy denies it?
- AAllow
- BDenyCorrect answer
- CConditionally allow
- DNot evaluated
Why this is the answer
An explicit deny in any authorization layer overrides an allow from another layer.
- Question 2Multiple Select
Which types of fine-grained access does Lake Formation support? (Select two)
- AColumn-levelCorrect answer
- BRow-levelCorrect answer
- CBucket-level
- DObject-level
Why this is the answer
Lake Formation supports column-level and row-level filtering for fine-grained access.
- Question 3Interactive
Refer to the bucket policy snippet. Which condition key enforces HTTPS?
- Aaws:SecureTransportCorrect answer
- Baws:SourceIp
- Cs3:x-amz-server-side-encryption
- Daws:SourceVpc
Why this is the answer
The condition key aws:SecureTransport with Bool condition enforces HTTPS.
- Question 4Single Choice
Which S3 server-side encryption option incurs no KMS charges?
- ASSE-S3Correct answer
- BSSE-KMS
- CSSE-C
- DClient-side encryption
Why this is the answer
SSE-S3 is free because S3 manages the keys without using KMS charges.
- Question 5Multiple Select
Which statements about SSE-KMS are true? (Select two)
- AIt uses envelope encryptionCorrect answer
- BIt charges per API requestCorrect answer
- CIt does not support encryption context
- DIt requires the client to manage keys
Why this is the answer
SSE-KMS uses envelope encryption and incurs KMS API charges for each object operation.
- Question 6Interactive
Examine the IAM policy. Which action is necessary to allow Lake Formation data access?
- Alakeformation:GetDataAccessCorrect answer
- Bathena:StartQueryExecution
- Cs3:GetObject
- Dkms:Decrypt
Why this is the answer
The IAM role must have lakeformation:GetDataAccess to retrieve temporary credentials from Lake Formation.
- Question 7Single Choice
Which de-identification technique is irreversible and used mainly for non-production data?
- AData maskingCorrect answer
- BTokenization
- CEncryption
- DHashing
Why this is the answer
Data masking is irreversible and used in non-production environments to protect sensitive data.
- Question 8Multiple Select
Which documents can you download from AWS Artifact? (Select two)
- ASOC reportsCorrect answer
- BPCI reportsCorrect answer
- CCloudTrail logs
- DS3 access logs
Why this is the answer
AWS Artifact provides on-demand access to SOC, PCI, ISO reports and agreements like BAA.
- Question 9Interactive
Refer to the AWS Config rule snippet. Which managed rule checks if S3 buckets have server-side encryption?
- As3-bucket-server-side-encryption-enabledCorrect answer
- Bs3-bucket-public-read-prohibited
- Cencrypted-volumes
- Dcloud-trail-enabled
Why this is the answer
The rule s3-bucket-server-side-encryption-enabled checks that S3 buckets have encryption enabled.
Data Ingestion and Orchestration
11 questions in this domain
- Question 10Single Choice
Which AWS service is primarily designed for batch ETL processing and includes a serverless Spark-based engine?
- AAWS GlueCorrect answer
- BAmazon Kinesis Data Streams
- CAmazon MSK
- DAWS DMS
Why this is the answer
AWS Glue is a serverless Spark-based ETL engine designed for batch processing. Kinesis Data Streams is for real-time streaming, MSK is managed Kafka, and DMS is for database migration.
- Question 11Multiple Select
Which of the following are features of Amazon Kinesis Data Streams? (Select two.)
- AData is split into shards with fixed write and read capacity.Correct answer
- BIt automatically batches and compresses records before delivery to S3.
- CRecords are ordered per shard for processing.Correct answer
- DIt is a fully managed service that requires no shard management.
Why this is the answer
Kinesis Data Streams uses shards with fixed throughput, and records are strictly ordered per shard. Automatic batching and compression are features of Kinesis Data Firehose. Shard management is required unless using On-Demand mode.
- Question 12Interactive
Examine the following code snippet for an AWS Glue job. What is the purpose of the `transformation_ctx` parameter used in the `create_dynamic_frame.from_catalog` call?
- AIt specifies the IAM role for the job run.
- BIt enables job bookmarks to track processed data for incremental processing.Correct answer
- CIt defines the output table name in the Data Catalog.
- DIt sets the number of Spark partitions for the DynamicFrame.
Why this is the answer
The `transformation_ctx` parameter is used by AWS Glue to uniquely identify the transformation and store bookmark state for incremental processing. It does not control IAM roles, output tables, or partition count.
- Question 13Single Choice
Which AWS service is used to migrate databases to AWS with minimal downtime by using change data capture (CDC)?
- AAWS Database Migration Service (DMS)Correct answer
- BAWS Glue
- CAmazon Kinesis Data Firehose
- DAmazon MSK
Why this is the answer
AWS DMS is specifically designed for database migration with CDC support, allowing near-zero downtime. Glue is for ETL, Kinesis Firehose for streaming delivery, and MSK for Kafka clusters.
- Question 14Multiple Select
Which of the following are valid use cases for Amazon Kinesis Data Firehose? (Select two.)
- ADelivering streaming data to Amazon S3 with near-real-time latency.Correct answer
- BProcessing millions of events per second with sub-millisecond latency and custom consumers.
- CIngesting log data from applications and delivering it to Amazon Redshift after transformation.Correct answer
- DMaintaining strict record ordering across all data for downstream applications.
Why this is the answer
Firehose is ideal for near-real-time delivery to S3 and Redshift with optional Lambda transformation. Sub-millisecond latency and strict ordering are not supported; those require Kinesis Data Streams or MSK.
- Question 15Interactive
Given the following AWS Step Functions state machine snippet, what state type is used to run multiple branches concurrently?
- ATask
- BParallelCorrect answer
- CMap
- DChoice
Why this is the answer
The `Parallel` state in Step Functions allows multiple branches to run concurrently. `Task` is for single actions, `Map` iterates over an array, and `Choice` implements conditional branching.
- Question 16Single Choice
What is the primary purpose of AWS Glue Workflows?
- ATo schedule and coordinate Glue jobs and crawlers within the Glue ecosystem.Correct answer
- BTo manage cross-account data ingestion from streaming sources.
- CTo provide a serverless Spark runtime for batch processing.
- DTo integrate with external API endpoints for data extraction.
Why this is the answer
Glue Workflows are designed for orchestrating Glue jobs and crawlers. They do not handle streaming ingestion, serve as compute engines, or integrate with external APIs natively.
- Question 17Multiple Select
Which of the following are key features of Amazon EventBridge? (Select two.)
- AEvent archiving with unlimited retention and replay capability.Correct answer
- BAutomatic schema discovery for custom events.Correct answer
- CBuilt-in support for database migration orchestration.
- DDirect integration with AWS Glue as a target without Lambda mediation.
Why this is the answer
EventBridge offers event archiving/replay and schema discovery (via Schema Registry). It does not natively orchestrate database migrations, and Glue is not a direct target; it typically requires Lambda or Step Functions as an intermediary.
- Question 18Interactive
The following EventBridge rule pattern is used to trigger a Lambda function. What must be configured on the Lambda side to allow EventBridge to invoke it?
- ACreate a resource-based policy on the Lambda function that grants `lambda:InvokeFunction` permission to the EventBridge service.Correct answer
- BAdd an environment variable to the Lambda function specifying the event bus name.
- CEnable versioning on the Lambda function to allow concurrent invocation.
- DConfigure a dead-letter queue (DLQ) in the Lambda function configuration.
Why this is the answer
EventBridge must have permission to invoke the Lambda function. This is done by adding a resource-based policy (also known as a Lambda permission) that allows EventBridge to call `lambda:InvokeFunction`. Environment variables, versioning, and DLQ are not required for basic invocation.
- Question 19Single Choice
What is the role of the AWS Glue Data Catalog in a data pipeline?
- AIt stores the actual data files in a compressed format.
- BIt provides a searchable index of metadata including table definitions, schemas, and partition information.Correct answer
- CIt executes ETL jobs and manages Spark clusters.
- DIt monitors pipeline performance and sends alerts.
Why this is the answer
The Data Catalog is a managed metadata repository that stores table schemas, partitions, and connection details. It does not store data files, run ETL jobs, or monitor pipelines.
- Question 20Multiple Select
Which of the following are prerequisites for AWS DMS to perform ongoing change data capture (CDC)? (Select two.)
- ABinary logging (for MySQL) or archived WAL (for PostgreSQL) must be enabled.Correct answer
- BThe source database must be hosted on Amazon RDS.
- CThe IAM role used by DMS must have permissions to read the source database schema.Correct answer
- DThe target database must be the same engine as the source.
Why this is the answer
CDC requires the source database to have binary logging (MySQL) or WAL archiving (PostgreSQL) enabled. DMS also needs the correct IAM permissions to read schema and tables. The source can be on-premises or EC2, and heterogeneous migrations are supported.
Data Processing and Transformation
11 questions in this domain
- Question 21Single Choice
Which AWS service provides a serverless Spark environment with integrated data cataloging and schema inference?
- AAWS GlueCorrect answer
- BAmazon EMR
- CAWS Lambda
- DAmazon Redshift Spectrum
Why this is the answer
AWS Glue is a serverless Spark service that provides a Data Catalog and automatic schema inference, suitable for ETL jobs with minimal operational overhead.
- Question 22Multiple Select
Which of the following are capabilities of Amazon EMR for data processing? (Select TWO)
- ASupports custom cluster configurations with Spark, Hive, and PrestoCorrect answer
- BProvides a serverless execution model with no cluster management
- CIncludes a built-in Data Catalog for schema management
- DAllows use of spot instances to reduce compute costsCorrect answer
Why this is the answer
Amazon EMR offers full control over Hadoop components (Spark, Hive, Presto) and can use spot instances for cost savings. It is not serverless and does not include a built-in Data Catalog (Glue is separate).
- Question 23Interactive
You are writing an external table definition in Amazon Redshift to query data in S3. Which SQL statement is correct for this purpose?
- ACREATE EXTERNAL TABLECorrect answer
- BCREATE TABLE AS SELECT
- CCREATE MATERIALIZED VIEW
- DCREATE FOREIGN TABLE
Why this is the answer
Redshift Spectrum uses CREATE EXTERNAL TABLE to define a schema for data in S3, allowing SQL queries without loading the data.
- Question 24Single Choice
Which SQL statement does Amazon Redshift use to create a new table from a query and store the results physically?
- ACREATE TABLE ASCorrect answer
- BSELECT INTO
- CCREATE VIEW AS
- DINSERT INTO SELECT
Why this is the answer
Redshift uses CREATE TABLE AS (CTAS) to create a new table and populate it with query results. SELECT INTO is not supported in Redshift.
- Question 25Multiple Select
Which of the following are required when using the SQL Query node in AWS Glue Studio? (Select TWO)
- AInput sources must be registered as temporary views before the SQL nodeCorrect answer
- BThe SQL node can only process data from Amazon Redshift
- CThe SQL node uses Spark SQL syntaxCorrect answer
- DThe SQL node exports results directly to a Redshift cluster
Why this is the answer
In Glue Studio, before using the SQL Query node, the DynamicFrames must be registered as temporary views. The node accepts Spark SQL, not standard SQL. It does not directly write to Redshift.
- Question 26Interactive
You want to convert a CSV table in Amazon Athena to Parquet format with partitioning. Which statement accomplishes this?
- ACREATE TABLE parquet_table WITH (format='PARQUET', partitioned_by=ARRAY['date']) AS SELECT * FROM csv_tableCorrect answer
- BCREATE TABLE parquet_table AS SELECT * FROM csv_table FORMAT PARQUET PARTITIONED BY (date)
- CALTER TABLE csv_table SET FORMAT PARQUET PARTITIONED BY (date)
- DCREATE EXTERNAL TABLE parquet_table STORED AS PARQUET PARTITIONED BY (date) AS SELECT * FROM csv_table
Why this is the answer
Athena's CTAS syntax uses WITH (format='PARQUET', partitioned_by=ARRAY['date']) to specify output format and partitioning. Other options use incorrect syntax or are not valid.
- Question 27Single Choice
What is AWS Deequ primarily used for in data processing pipelines?
- AAutomating data quality checks on large datasets using SparkCorrect answer
- BProviding a visual interface for data profiling and cleansing
- COrchestrating complex ETL workflows with state machines
- DMonitoring real-time streaming data for anomalies
Why this is the answer
Deequ is an open-source library for defining and running data quality constraints on Spark DataFrames, integrated with AWS Glue. DataBrew is the visual profiling service.
- Question 28Multiple Select
Which of the following are examples of custom validation logic that complement Deequ? (Select TWO)
- AChecking referential integrity across multiple tablesCorrect answer
- BValidating that a column contains only numeric values
- CDetecting schema drift by comparing DataFrame schema to a canonical schemaCorrect answer
- DEnsuring a column has no missing values
Why this is the answer
Deequ handles standard checks like completeness and data types. Cross-table referential integrity and schema drift require custom logic because they involve multiple sources or are not covered by Deequ's built-in constraints.
- Question 29Interactive
You need to check schema drift in a Glue job. Which code snippet correctly compares the DataFrame schema with a known structure?
- Adf.schema == expected_schemaCorrect answer
- Bdf.columns == expected_columns
- Cdf.dtypes == expected_dtypes
- Ddf.printSchema() == expected_schema
Why this is the answer
In PySpark, df.schema returns a StructType that can be compared to an expected StructType. The other options do not provide full type information or are not direct comparisons.
- Question 30Single Choice
What is the default value of spark.sql.shuffle.partitions in Apache Spark?
- A200Correct answer
- B100
- C50
- D400
Why this is the answer
The default value for spark.sql.shuffle.partitions is 200. This can be adjusted based on data size and cluster resources.
- Question 31Multiple Select
Which of the following are recommended partitioning strategies for Amazon S3 data lakes? (Select TWO)
- APartition by low-cardinality columns like date or regionCorrect answer
- BPartition by high-cardinality columns like user_id
- CUse static partitioning with Hive-style directory pathsCorrect answer
- DAvoid using any partitioning for optimal performance
Why this is the answer
Low-cardinality columns (e.g., date, region) are ideal for partitioning because they limit the number of directories and enable efficient pruning. Static partitioning with Hive-style paths is a common pattern. High-cardinality columns lead to too many small files.
Data Storage and Data Management
11 questions in this domain
- Question 32Single Choice
Which Amazon S3 storage class is best for frequently accessed data?
- AS3 StandardCorrect answer
- BS3 Glacier Deep Archive
- CS3 One Zone-IA
- DS3 Glacier Flexible Retrieval
Why this is the answer
S3 Standard provides low latency and high throughput for frequently accessed data.
- Question 33Multiple Select
Which S3 storage classes are suitable for archival data? (Select two.)
- AS3 Glacier Instant RetrievalCorrect answer
- BS3 Glacier Deep ArchiveCorrect answer
- CS3 Standard-IA
- DS3 One Zone-IA
Why this is the answer
Glacier Instant Retrieval and Glacier Deep Archive are archival storage classes.
- Question 34Interactive
Analyze the DDL and choose the correct distribution style.
- AKEY distribution on customer_idCorrect answer
- BEVEN distribution
- CALL distribution
- DAUTO distribution
Why this is the answer
For large tables joined on a column, KEY distribution collocates related rows.
- Question 35Single Choice
Which DynamoDB partition key design prevents hot partitions?
- AUse high-cardinality attributesCorrect answer
- BUse sequential numbers
- CUse a single attribute only
- DUse date as partition key
Why this is the answer
High-cardinality keys distribute writes evenly across partitions, avoiding hotspots.
- Question 36Multiple Select
Which statements about Amazon Aurora are true? (Select two.)
- ASupports up to 15 read replicasCorrect answer
- BStorage auto-scales up to 128 TBCorrect answer
- CWrite instance can auto-scale vertically
- DOnly supports MySQL compatibility
Why this is the answer
Aurora scales storage to 128 TB and provides up to 15 low-lag read replicas.
- Question 37Interactive
Examine the data sample and choose the best file format.
- AParquetCorrect answer
- BAvro
- CCSV
- DJSON
Why this is the answer
Parquet is columnar and efficient for analytical queries that select few columns.
- Question 38Single Choice
When should you use a transient EMR cluster?
- ARun a batch job and terminateCorrect answer
- BRun interactive queries all day
- CStore persistent data in HDFS
- DReplace a production database
Why this is the answer
Transient clusters are cost-effective for batch ETL jobs that terminate after completion.
- Question 39Multiple Select
Which two actions can an S3 Lifecycle policy perform? (Select two.)
- ATransition objects to GlacierCorrect answer
- BExpire objects after a periodCorrect answer
- CReplicate objects across regions
- DEncrypt objects at rest
Why this is the answer
Lifecycle rules can transition storage classes and expire objects. Replication and encryption are separate features.
- Question 40Interactive
Identify the medallion layer for raw, immutable data.
- ABronzeCorrect answer
- BSilver
- CGold
- DPlatinum
Why this is the answer
Bronze layer stores raw ingested data in original format, preserving immutability.
- Question 41Single Choice
Which compression codec is best for write-heavy workloads?
- ASnappyCorrect answer
- BGzip
- CBzip2
- DZstd
Why this is the answer
Snappy offers fast compression and decompression, ideal for write-intensive ETL pipelines.
- Question 42Multiple Select
Which benefits does a columnar file format provide? (Select two.)
- ABetter compression ratiosCorrect answer
- BPredicate pushdown supportCorrect answer
- CFaster single-record inserts
- DSimpler schema evolution
Why this is the answer
Columnar formats compress well and enable predicate pushdown to skip irrelevant data.
Frequently asked questions
Everything about the AWS Certified Data Engineer – Associate question bank and how to use it.