AWS Certified Data Engineer Associate DEA-C01 Practice Questions with Explanations

156 questions · Cloud Computing · by Study Foxy

Study these questions free

Questions

  1. A Kinesis producer gets ProvisionedThroughputExceededException on one shard while others sit idle. What is the most likely cause?

    • A low-cardinality partition key sending most records to one shard
    • Too many consumer applications reading the stream
    • A retention period that is set too long
    • Server-side encryption enabled on the stream

  2. Five consumer apps read the same Kinesis stream and are competing for the 2 MB/s read limit per shard. What gives each consumer its own throughput?

    • Enhanced fan-out
    • A longer retention period
    • On-demand capacity mode only
    • A larger batch size in each consumer

  3. A bug corrupted three days of processed output. The team wants to reprocess the raw events from the stream. What makes this possible?

    • A longer retention period
    • Enhanced fan-out
    • Firehose buffering
    • Lambda reserved concurrency

  4. A company must replicate an on-premises Oracle database to Amazon S3 with an initial copy and then ongoing changes. Which service fits?

    • AWS DMS with full load plus CDC
    • AWS DataSync with a daily schedule
    • Amazon AppFlow with an Oracle connector
    • AWS Transfer Family with SFTP

  5. Sales data must be pulled from Salesforce into S3 every hour with no custom code. Which service fits?

    • Amazon AppFlow
    • AWS DMS
    • AWS DataSync
    • Amazon Data Firehose

  6. A Glue crawler must run every night at 1 a.m. What is the simplest way to schedule it?

    • A cron-based schedule on the crawler
    • An S3 lifecycle rule on the bucket
    • A CloudTrail trail on the Glue API
    • A Kinesis stream that triggers the crawler

  7. A pipeline should start as soon as a new file lands in an S3 prefix. Which trigger fits?

    • An S3 Event Notification or an EventBridge rule
    • A Glue crawler that runs once a week
    • A Redshift materialized view refresh on a 5-minute schedule
    • An AWS Config rule on the bucket

  8. A Lambda function processing a Kinesis stream falls behind. Records within each shard can be processed in parallel by key. What increases throughput?

    • Raise the parallelization factor
    • Lower the batch size to 1
    • Shorten the stream's retention period to one hour
    • Turn off enhanced fan-out

  9. One bad record in a Kinesis batch keeps failing the Lambda function and blocks the shard. Which setting isolates it?

    • Bisect batch on function error
    • Increase the parallelization factor
    • Lower the stream's shard count
    • Disable the event source mapping

  10. An AWS Glue job must connect to an RDS database through a JDBC connection inside a VPC, but it times out. What is a commonly missed requirement?

    • A self-referencing inbound rule on the security group
    • A public IP address on the RDS instance
    • An internet gateway route added for the Glue job's subnet
    • Encryption disabled on the database

  11. A third-party database only accepts connections from known IP addresses. How can AWS jobs connect from a stable address?

    • Route through a NAT gateway and allowlist its Elastic IP address
    • Give each Glue worker its own public IP address
    • Use a gateway VPC endpoint to reach the database
    • Ask the vendor to allow every AWS IP range

  12. A batch job writing to DynamoDB is throttled. Which client-side approach handles this correctly?

    • Retries with exponential backoff and jitter
    • Immediate retries in a tight loop
    • Dropping failed writes silently
    • Switching all of the writes over to Scan calls

  13. Which describes fan-out in streaming data distribution?

    • One stream delivered to many consumers
    • Many producers writing to one stream
    • Merging several streams into one table
    • Compressing records before sending them

  14. A stream processor computes a running total per customer across events. What kind of processing is this?

    • Stateful
    • Stateless
    • Idempotent only
    • Batch-only

  15. Which service reads item-level changes from a DynamoDB table in near real time?

    • DynamoDB Streams
    • DynamoDB Accelerator (DAX)
    • DynamoDB global tables only
    • DynamoDB TTL

  16. Kinesis Data Streams lets multiple consumer applications read the same records independently.

    True or false

  17. Which help a Kinesis Data Streams pipeline handle more write throughput? (Select TWO.)

    • Add shards or use on-demand capacity mode
    • Use a higher-cardinality partition key
    • Lower the retention period to 1 hour
    • Disable server-side encryption on the stream

  18. Analysts query CSV logs in S3 with Athena, and costs are high. What reduces data scanned the most?

    • Convert to partitioned Parquet
    • Add more CSV columns
    • Gzip the CSV files only
    • Move the CSV files to S3 Glacier Flexible Retrieval

  19. Streaming JSON must land in S3 as Parquet with no code to manage. Which service fits?

    • Amazon Data Firehose with format conversion
    • Kinesis Data Streams with a Lambda consumer
    • AWS DMS with a JSON target
    • Amazon AppFlow with a schedule

  20. A nightly Glue job keeps reprocessing all files in a prefix instead of only new ones. What should be enabled?

    • Job bookmarks
    • Glue Flex execution
    • Auto scaling workers
    • A Glue Schema Registry

  21. A non-urgent nightly Glue job should run at lower cost and can tolerate variable start times. What should be used?

    • Glue Flex execution
    • G.2X workers with auto scaling
    • Provisioned concurrency
    • Glue Streaming jobs

  22. A long-running Spark job on EMR needs to cut cost while keeping core HDFS nodes stable. What should run on Spot Instances?

    • Task nodes
    • The primary node
    • Core nodes only
    • Every node in the cluster

  23. A team wants to run Spark jobs without managing clusters at all. Which option fits?

    • Amazon EMR Serverless for Spark and Hive jobs
    • EMR on EC2 with a long-running cluster
    • Amazon RDS with read replicas
    • Amazon MemoryDB for Redis

  24. A Spark job slows down because one join key holds most of the rows. What technique helps?

    • Salt the skewed key to spread its rows across partitions
    • Increase the S3 retention period
    • Convert the data to CSV
    • Reduce the number of executors

  25. An ETL step must read from both a Redshift table and an on-premises SQL Server database. Which connectivity standard does Glue use for these sources?

    • JDBC connections
    • SMB file shares
    • SFTP endpoints
    • MQTT topics

  26. Other teams need to query a Redshift cluster over HTTPS without managing persistent database connections. What should they use?

    • The Redshift Data API, which runs SQL asynchronously
    • A JDBC connection pool on EC2
    • An S3 presigned URL to the cluster
    • A Kinesis stream into the cluster

  27. A pipeline must extract product attributes from free-text supplier descriptions during ETL. Which approach uses an LLM?

    • Call a Bedrock foundation model
    • Run a Glue crawler on the text files
    • Store the text in DynamoDB with TTL
    • Compress the text with Snappy

  28. A Glue Spark job fails with out-of-memory errors on large partitions. What is a reasonable first fix?

    • Use larger worker types
    • Switch the output to CSV
    • Disable job bookmarks
    • Lower the number of retries

  29. Which describes the three Vs of big data?

    • Volume, velocity, and variety
    • Validity, versioning, and value
    • Velocity, visibility, and vectors
    • Volume, virtualization, and VPCs

  30. A web app needs a REST endpoint that returns aggregated sales from a data store. Which combination fits?

    • API Gateway in front of Lambda
    • Kinesis Data Streams in front of S3
    • AWS DMS in front of Redshift
    • Glue crawlers in front of Athena

  31. Which service runs real-time stream processing with windowed aggregations using Apache Flink?

    • Amazon Managed Service for Apache Flink
    • Amazon AppFlow
    • AWS Glue DataBrew
    • Amazon Data Firehose with buffering

  32. A team already writes its workflows as Apache Airflow DAGs and wants a managed environment. Which service fits?

    • Amazon MWAA
    • AWS Glue DataBrew
    • Amazon AppFlow
    • AWS Batch

  33. A serverless pipeline must run Glue jobs and Lambda functions in order, with retries and error branches. Which service fits?

    • AWS Step Functions
    • Amazon SQS
    • Amazon SNS
    • Amazon EventBridge Scheduler

  34. A Step Functions workflow must process 1,000 files in parallel with the same steps. Which state fits?

    • Map state
    • Choice state
    • Wait state
    • Pass state

  35. Several Glue crawlers and jobs must run in a dependency chain entirely inside Glue. Which feature fits?

    • Glue workflows with triggers
    • Glue Schema Registry
    • Glue Data Catalog table versions
    • Glue DataBrew recipes

  36. The data team must get an alert when a nightly pipeline fails. What is a common approach?

    • Route the pipeline's failure events to an SNS topic for alerts
    • A Glue crawler on the logs bucket
    • An S3 lifecycle rule on the outputs
    • A Redshift materialized view

  37. Which design improves a pipeline's fault tolerance?

    • Idempotent steps that can be safely retried after a failure
    • One huge step that does everything
    • Deleting raw data after loading
    • Running all jobs on one EC2 instance

  38. A team wants a pipeline with no servers to patch that scales automatically. Which set of services fits?

    • Lambda, Step Functions, and Glue
    • EC2, cron, and a self-managed database
    • EMR on EC2 with long-running clusters
    • On-premises Hadoop with a VPN

  39. Which tool packages and deploys Lambda functions, Step Functions, and DynamoDB tables for a serverless pipeline from one template?

    • AWS SAM
    • AWS DataSync
    • Amazon AppFlow
    • AWS Glue DataBrew

  40. A Lambda function needs to read large reference files shared across invocations and other compute. What can it mount?

    • An Amazon EFS file system
    • An EBS volume
    • An S3 Glacier vault
    • An instance store volume from the Lambda host

  41. A Lambda function in a pipeline is throttled during bursts, starving other functions. What limits it while reserving capacity?

    • Reserved concurrency
    • A larger memory setting
    • A longer timeout
    • A Lambda layer

  42. Which practice brings software engineering discipline to data pipelines?

    • Version control, tests, and CI/CD
    • Editing production jobs in the console
    • Skipping tests to deploy faster
    • Keeping scripts only on one laptop

  43. A team wants to define its data infrastructure in Python and deploy it repeatably. Which tool fits?

    • AWS CDK
    • AWS DataSync
    • Amazon Macie
    • AWS Glue DataBrew

  44. Which data structure is best for representing relationships such as which accounts share devices?

    • A graph
    • A queue
    • A flat array
    • A stack

  45. What is distributed computing in the context of data engineering?

    • Splitting work across many machines
    • Running all jobs one after another on one host
    • Storing all data in a single file
    • Copying data to every AWS Region

  46. Which change usually reduces runtime for a Python transformation that loops row by row over millions of records?

    • Use vectorized or set-based operations
    • Add more print statements for logging each row
    • Convert the data to XML first
    • Process one file per Lambda per day

  47. Firehose delivers files to S3 that are too small and too frequent. Which setting produces fewer, larger files?

    • Increase the buffer size and buffer interval on the stream
    • Lower the buffer interval to zero seconds
    • Turn on enhanced fan-out for the stream
    • Add a second delivery stream to the same bucket

  48. Records in a Kafka-based pipeline must be ingested into AWS with a fully managed Kafka service. Which fits?

    • Amazon MSK
    • Amazon MQ
    • Amazon AppFlow
    • AWS DataSync

  49. Amazon Data Firehose is designed for sub-second, record-by-record delivery to its destinations.

    True or false

  50. A Firehose stream must enrich each record with a lookup before it's written to S3. Which feature fits?

    • A Lambda transformation on the stream
    • A Glue crawler on the destination
    • An S3 lifecycle rule
    • A Redshift materialized view

Showing 50 of 156. Open the quiz to study them all.