DEA-C01 Data Operations and Support Practice Test 12 — 15 Questions

Question 1

A data engineer is running a Glue ETL job that reads from a JDBC source and writes to S3 in Parquet format. The job is slow and the engineer notices that the number of DPUs used is low. What can be done to improve performance?

Choose an answer to begin — your selection is scored in the full session.

Accepted Answer

Increase the number of workers (MaxCapacity) in the job configuration.. Increasing the number of workers (MaxCapacity) directly increases the parallelism of the Glue ETL job. Since the DPU usage is low, the job is not fully utilizing available resources; adding workers allows more concurrent processing, thus improving performance. Option A: Disabling job bookmarks avoids reading metadata but does not address low DPU usage and may cause reprocessing. Option B: Push-down predicates filter data at the source, reducing data volume, but the primary bottleneck here is parallelism, not data volume. Option D: Changing output to CSV would increase overhead and worsen performance because Parquet is columnar and compressed, while CSV is not.

Answer

Disable job bookmarks to avoid reading metadata.

Answer

Use push-down predicates to filter data at the source.

Answer

Change the output format to CSV to reduce CPU overhead.

Question 2

A data engineer is setting up a Redshift cluster and needs to ensure high availability. Which TWO actions should be taken?

Accepted Answer

Configure cross-region snapshot copy.. For high availability in Amazon Redshift, you need to protect against data loss and downtime. Option B (configure cross-region snapshot copy) ensures that snapshots are replicated to another region, providing disaster recovery and high availability in case of a regional outage. Option E (deploy a multi-node cluster with at least two compute nodes) provides node-level redundancy; if one node fails, the workload can be redistributed. Option C (enable automatic replication across Availability Zones) is not a feature of Amazon Redshift; Redshift does not support automatic cross-AZ replication for compute nodes. Option A (concurrency scaling) improves query concurrency, not availability. Option D (single-node cluster) offers no redundancy.

Answer

Enable concurrency scaling.

Answer

Enable automatic replication across Availability Zones.

Answer

Use a single-node cluster to reduce complexity.

Question 3

A company is using Amazon Athena to query data in an S3 bucket. Queries are failing with the error 'HIVE_PATH_ALREADY_EXISTS'. The data is partitioned by year, month, day. What is the MOST likely cause?

Accepted Answer

A partition was manually added to the Glue Data Catalog that already exists. The error 'HIVE_PATH_ALREADY_EXISTS' occurs in Athena when attempting to add a partition (via ALTER TABLE ADD PARTITION or MSCK REPAIR TABLE) that already exists in the Glue Data Catalog. Option B (inconsistent data format) would cause schema mismatch errors like 'HIVE_PARTITION_SCHEMA_MISMATCH', not this error. Option C (empty S3 location) would not cause this error; queries might succeed but return no results. Option D (lack of s3:ListBucket permission) would cause permission errors like 'Access Denied'.

Answer

The data format in the partition is inconsistent with the table schema

Answer

The S3 location for the partition is empty

Answer

The IAM role used by Athena lacks s3:ListBucket permission on the bucket

Question 4

A company is using Amazon EMR to run Spark jobs. The jobs are failing due to memory issues. Which THREE configurations can help mitigate out-of-memory errors?

Accepted Answer

Tune Spark memory configurations like spark.executor.memory and spark.memory.fraction.. The correct options are C, D, and E. Tuning Spark memory configurations (C) such as spark.executor.memory and spark.memory.fraction directly controls memory allocation within executors. Increasing the instance type to one with more memory per node (D) provides additional physical memory for Spark workloads. Enabling Spark dynamic allocation (E) allows the cluster to automatically adjust the number of executors based on workload, which helps prevent memory pressure from over-allocation. Option A is incorrect because instance store volumes are used for temporary data storage (e.g., shuffle spills) but do not address memory constraints; they may help with disk I/O but not OOM errors. Option B is incorrect because increasing vCPUs increases parallelism, which can actually worsen memory contention if each task consumes significant memory, potentially leading to more OOM errors.

Answer

Configure instance store volumes for intermediate shuffle data.

Answer

Use instances with more vCPUs to process more tasks in parallel.

Question 5

A data engineer is designing a data pipeline that ingests millions of small JSON files (1-10 KB each) from an S3 bucket into Amazon Redshift. The current approach uses a Lambda function triggered by S3 events to call the Redshift COPY command for each file. This is causing high latency and throttling. Which alternative is MOST cost-effective and efficient?

Accepted Answer

Use Amazon Kinesis Data Firehose to buffer and write larger files to S3, then use a scheduled COPY command. Amazon Kinesis Data Firehose can buffer the incoming small JSON files from S3 (via S3 event notifications) and write larger aggregated files to S3. A scheduled COPY command then efficiently loads these larger files into Amazon Redshift, reducing the number of COPY operations and avoiding Lambda throttling. This approach is cost-effective as Firehose charges only for data volume processed, and it eliminates the need for custom batching logic. Other options either process files individually (A, C) or incur higher costs with AWS Glue (D).

Answer

Use Amazon Kinesis Data Streams and a consumer to batch files before COPY

Answer

Increase the Lambda concurrency limit and memory

Answer

Use AWS Glue to merge files into larger Parquet files before loading

Question 6

A data engineer needs to monitor the performance of an RDS for PostgreSQL database. Which THREE CloudWatch metrics are most useful for this purpose?

Accepted Answer

CPUUtilization. CPUUtilization is a critical metric for monitoring RDS for PostgreSQL because high CPU usage can indicate inefficient queries, insufficient instance size, or contention. Sustained high CPU can lead to performance degradation and increased query latency, making it essential for capacity planning and troubleshooting.

Answer

FreeStorageSpace

Answer

NetworkThroughput

Question 7

A company runs an Amazon EMR cluster that processes data from S3 and writes results back to S3. The cluster uses Spot Instances for task nodes. Some tasks are failing due to Spot Instance interruptions. What is the BEST way to handle this without manual intervention?

Accepted Answer

Enable automatic node replacement in the EMR cluster. Amazon EMR's automatic node replacement feature automatically detects when a Spot Instance is interrupted and launches a replacement instance, ensuring the cluster continues processing without manual intervention. Option B is incorrect because manually relaunching the cluster requires human intervention and is not automated. Option C is incorrect because while checkpointing to S3 can help recover data after failures, it does not automatically replace interrupted instances. Option D is incorrect because using only On-Demand instances eliminates cost savings from Spot Instances and does not address the interruption handling problem—it avoids interruptions entirely by not using Spot Instances, but the best approach is to handle interruptions automatically with automatic node replacement.

Answer

Manually relaunch the cluster after failures

Answer

Configure the application to checkpoint to S3 every few minutes

Answer

Use only On-Demand instances for task nodes

Question 8

Refer to the exhibit. An IAM policy is attached to an IAM role used by an application. The application needs to read objects from 'my-bucket' that have the tag 'classification=public'. The application account is 123456789012. However, the application is getting 'Access Denied' errors. What is the most likely reason?

Accepted Answer

The object being accessed does not have the tag 'classification=public'.. The Allow statement grants s3:GetObject only when the request has a condition that the object tag 'classification' equals 'public' (using StringEquals). If the object being accessed does not have this tag, the Allow condition is not satisfied, and the request is implicitly denied. The Deny statement does not apply because it only denies when the source account is NOT 123456789012, and the request does come from that account. Thus, the most likely cause is that the object's tag does not match 'classification=public'.

Answer

The Deny statement uses StringNotEquals, which incorrectly denies the application account.

Answer

The policy does not grant s3:ListBucket permission, so the application cannot list objects.

Answer

The Deny statement blocks all access from accounts other than 123456789012, but the application is in that account.

Question 9

A company uses AWS Glue DataBrew to clean and transform data. A data engineer notices that a DataBrew recipe step that should remove duplicates is not working as expected. The dataset has millions of rows. What is the MOST likely reason?

Accepted Answer

The dataset is not sorted by the columns used for deduplication. DataBrew's 'Remove duplicate rows' step identifies duplicates by comparing each row to the previous row in the dataset. If the data is not sorted by the key columns, duplicate rows may not be adjacent and thus will not be removed, causing the step to appear ineffective. Option A: The number of files does not directly affect deduplication logic. Option B: Null values may be treated as distinct, but the most likely issue is lack of sorting. Option D: DataBrew projects can use sampling for preview, but recipes are applied to the full dataset when run, so sampling does not prevent deduplication.

Answer

The data source is an S3 bucket with a large number of files

Answer

The dataset contains null values in the key columns

Answer

The DataBrew project is using a sampling of the data

Question 10

Refer to the exhibit. A data engineer sees this error in CloudWatch Logs from an AWS Glue ETL job. The job reads from an S3 location that contains both .parquet and .csv files. What is the most likely cause?

Accepted Answer

The job is reading a CSV file that was incorrectly placed in the directory with .parquet extension.. The error indicates that the job encountered an object that is not a valid Parquet file. Since the S3 location contains both .parquet and .csv files, the Glue job likely attempted to read a CSV file as if it were Parquet, causing the invalid Parquet error. Option C is correct because the CSV file was incorrectly placed in the directory with a .parquet extension (or the job's schema inference expects all files to be Parquet). Option A is incorrect because the error is about format, not a missing file. Option B is incorrect because the error is about the object's format, not a permissions issue. Option D is incorrect because insufficient memory would typically cause out-of-memory or capacity errors, not an invalid Parquet error.

Answer

The S3 object was deleted during the job execution.

Answer

The IAM role does not have permission to read the S3 object.

Answer

The Glue job does not have enough memory to parse the Parquet file.

Question 11

A data engineer needs to monitor the number of records processed by an Amazon Kinesis Data Analytics application and trigger an alarm if the count drops below a threshold over 5 minutes. Which CloudWatch metric should be used?

Accepted Answer

millisBehindLatest (from KinesisDataAnalytics). 'millisBehindLatest' is a metric published by KinesisDataAnalytics that indicates how far behind the application is in processing records. Monitoring this metric allows the engineer to set an alarm when the number of records processed drops, as a high millisBehindLatest value indicates the application is falling behind. Option B (IncomingRecords) is from Kinesis Streams, not KinesisDataAnalytics. Option C (DPUCount) is a Glue metric. Option D (IncomingBytes) is a Kinesis Firehose metric.

Answer

IncomingRecords (from Kinesis Streams)

Answer

DPUCount (from Glue)

Answer

IncomingBytes (from Kinesis Firehose)

Question 12

Refer to the exhibit. A CloudFormation template is used to create a DynamoDB table. After creation, a data engineer wants to restore the table to a point in time from 3 hours ago. Which action is required?

Accepted Answer

Use the AWS CLI or Console to initiate a point-in-time restore specifying the desired timestamp.. To restore a DynamoDB table to a point in time, the table must have point-in-time recovery (PITR) enabled. If it is enabled, you can use the AWS CLI or Console to initiate a point-in-time restore by specifying the desired timestamp. Option A is incorrect because a manual backup is not required; PITR uses continuous backups. Option B is incorrect because AWS Backup is not necessary; DynamoDB PITR is sufficient. Option C is incorrect because PITR does not require any on-demand backups; it relies on continuous backups.

Answer

Create a manual backup of the table first.

Answer

Enable AWS Backup to schedule automatic backups.

Answer

Ensure the table has at least one on-demand backup.

Question 13

Which TWO actions should a data engineer take to optimize Amazon S3 query performance for Amazon Athena when dealing with large Parquet files? (Choose 2.)

Accepted Answer

Optimize file sizes to be around 64 MB to 256 MB. The correct answers are D and E. Optimizing file sizes to be around 64 MB to 256 MB (D) reduces the overhead of Athena reading many small files or processing overly large files, balancing parallelism and efficiency. Partitioning data by frequently filtered columns (E) allows Athena to prune partitions and scan less data, improving query performance. Option A (single large file) hinders parallelism and query performance. Option B (GZIP compression) is already supported by Parquet and does not inherently optimize query performance beyond the file size considerations. Option C (many small files) increases metadata overhead and degrades performance.

Answer

Store data in a single large file without partitioning

Answer

Use GZIP compression on the Parquet files

Answer

Split large files into many small files

Question 14

A data engineer is troubleshooting a nightly ETL job that extracts data from an Amazon RDS MySQL instance and loads it into an Amazon S3 bucket in Parquet format. The job runs on an Amazon EMR cluster and has been failing with the error 'Access Denied' when writing to S3. The IAM role attached to the EMR cluster has permissions for S3 PutObject. What is the MOST likely cause?

Accepted Answer

The S3 bucket policy denies access to the EMR cluster's IAM role.. S3 bucket policies can override IAM permissions; if the bucket policy denies access from the EMR cluster, the write will fail even with IAM allow. Option A is wrong because KMS permissions are needed only if the bucket uses SSE-KMS, which is not indicated. Option B is wrong because S3 Lifecycle rules do not affect write permissions. Option C is wrong because EMR cluster termination would cause a different error.

Answer

The S3 bucket uses SSE-KMS encryption and the EMR role lacks kms:GenerateDataKey permission.

Answer

The S3 bucket has a Lifecycle rule that expires objects too quickly.

Answer

The EMR cluster was terminated before the write operation completed.

Question 15

Which THREE considerations are important when designing a data pipeline that uses AWS Glue to process streaming data from Amazon Kinesis Data Streams? (Choose 3.)

Accepted Answer

Set the number of Glue workers to match the number of shards for optimal parallelism. A, B, and C are correct. A: Setting the number of Glue workers to match the number of shards ensures optimal parallelism and resource utilization. B: Ensuring sufficient shards in Kinesis Data Streams is essential to handle the expected record rate and avoid throttling. C: Configuring checkpointing in AWS Glue streaming jobs prevents data loss and allows resumption from the last checkpoint on failure. D (batch window) is not appropriate for streaming pipelines as it introduces latency and is more suited for batch processing. E (converting to Avro) is a general data optimization but not a critical consideration specifically for Glue streaming with Kinesis.

Answer

Use batch window to accumulate data before processing

Answer

Convert data to Avro format for better compression