92 free AWS DEA-C01 practice questions from the CertStash pack of 366, each with the correct answer and a full explanation. The rule is the same for every exam: ten free on the product page, 92 free here, all 366 in the pack, and each step includes the one before it.
Get all 366 questions (US$39) · Download the first 10 as a PDF · Read the study guide
Question 1
A data engineer is configuring an AWS Glue job to read data from an Amazon S3 bucket. The data engineer has set up the necessary AWS Glue connection details and an associated IAM role. However, when the data engineer attempts to run the AWS Glue job, the data engineer receives an error message that indicates that there are problems with the Amazon S3 VPC gateway endpoint. The data engineer must resolve the error and connect the AWS Glue job to the S3 bucket. Which solution will meet this requirement?
Show answer and explanation
Correct answer: D. Verify that the VPC's route table includes inbound and outbound routes for the Amazon S3 VPC gateway endpoint.
When AWS Glue attempts to access S3 through a VPC gateway endpoint, the VPC route table must contain the appropriate routes to enable traffic to reach the endpoint. The route table needs both inbound and outbound routes configured for the S3 VPC gateway endpoint so that the Glue job's traffic is properly directed through the endpoint. This is a routing configuration issue at the VPC level, not a security group, bucket policy, or application code issue.
Why the other options are wrong
- A. Security groups control traffic to EC2 instances; VPC gateway endpoints do not receive inbound traffic from security groups in the same way.
- B. An S3 bucket policy grants permissions but does not resolve VPC endpoint routing issues that prevent connectivity.
- C. The Glue connection details do not need a fully qualified domain name for S3 VPC gateway endpoints; routing tables are the issue.
Question 2
A retail company has a customer data hub in an Amazon S3 bucket. Employees from many countries use the data hub to support company-wide analytics. A governance team must ensure that the company's data analysts can access data only for customers who are within the same country as the analysts. Which solution will meet these requirements with the LEAST operational effort?
Show answer and explanation
Correct answer: B. Register the S3 bucket as a data lake location in AWS Lake Formation. Use the Lake Formation row-level security features to enforce the company's access policies.
AWS Lake Formation provides built-in row-level security features that allow enforcement of access policies based on attributes like country. By registering the S3 bucket as a data lake location in Lake Formation and configuring row-level security, analysts can access the same table but only see rows relevant to their country. This requires minimal operational effort compared to creating separate tables, managing multiple regions, or building custom Redshift views and IAM roles.
Why the other options are wrong
- A. Creating separate tables for each country is operationally intensive and does not scale well as countries are added.
- C. Moving data to multiple regions introduces significant complexity, cost, and operational overhead for a data governance problem.
- D. Using Amazon Redshift with separate views and IAM roles requires more manual management and operational effort than Lake Formation's built-in security features.
Question 3
A media company wants to improve a system that recommends media content to customer based on user behavior and preferences. To improve the recommendation system, the company needs to incorporate insights from third-party datasets into the company's existing analytics platform. The company wants to minimize the effort and time required to incorporate third-party datasets. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: A. Use API calls to access and integrate third-party datasets from AWS Data Exchange.
AWS Data Exchange is purpose-built for accessing and integrating third-party datasets with minimal effort. It provides pre-packaged datasets that can be accessed via API calls and easily integrated into existing analytics platforms. This is the most straightforward solution with the least operational overhead compared to alternatives.
Why the other options are wrong
- B. AWS DataSync is designed for data transfer and synchronization between storage systems, not for accessing third-party datasets.
- C. Amazon Kinesis Data Streams is a real-time streaming service, not a third-party dataset source; AWS CodeCommit is for version control.
- D. Kinesis Data Streams is not designed for dataset integration; Amazon ECR is a container registry, not a third-party dataset source.
Question 4
A financial company wants to implement a data mesh. The data mesh must support centralized data governance, data analysis, and data access control. The company has decided to use AWS Glue for data catalogs and extract, transform, and load (ETL) operations. Which combination of AWS services will implement a data mesh? (Choose two.)
Show answer and explanation
Correct answer: B, E. Use Amazon S3 for data storage. Use Amazon Athena for data analysis; Use AWS Lake Formation for centralized data governance and access control.
A data mesh implementation requires data storage, data analysis capabilities, and centralized governance with access control. Amazon S3 provides scalable data storage for the mesh, Amazon Athena enables serverless data analysis, and AWS Lake Formation provides centralized data governance and fine-grained access control. Together, these services support all requirements of a data mesh architecture.
Why the other options are wrong
- A. Amazon Aurora is a relational database suitable for transactional workloads, not a data mesh storage pattern; Redshift requires cluster management overhead.
- C. AWS Glue DataBrew is a data preparation tool, not a governance and access control solution.
- D. Amazon RDS is a relational database not suited for data mesh scale; EMR requires more operational overhead than Athena.
Question 5
A data engineer maintains custom Python scripts that perform a data formatting process that many AWS Lambda functions use. When the data engineer needs to modify the Python scripts, the data engineer must manually update all the Lambda functions. The data engineer requires a less manual way to update the Lambda functions. Which solution will meet this requirement?
Show answer and explanation
Correct answer: B. Package the custom Python scripts into Lambda layers. Apply the Lambda layers to the Lambda functions.
Lambda layers allow packaging of custom Python scripts once and applying them to multiple Lambda functions. When the scripts are updated, the layer is updated once and all Lambda functions automatically use the new version. This eliminates manual updates to each individual function, providing a centralized, maintainable solution.
Why the other options are wrong
- A. Storing pointers in the execution context requires manual management and does not provide centralized versioning.
- C. Using environment variables to store pointers still requires manual updates to each Lambda function configuration.
- D. Lambda aliases are used for version management of the functions themselves, not for sharing code across functions.
Question 6
A company created an extract, transform, and load (ETL) data pipeline in AWS Glue. A data engineer must crawl a table that is in Microsoft SQL Server. The data engineer needs to extract, transform, and load the output of the crawl to an Amazon S3 bucket. The data engineer also must orchestrate the data pipeline. Which AWS service or feature will meet these requirements MOST cost-effectively?
Show answer and explanation
Correct answer: B. AWS Glue workflows
AWS Glue workflows are purpose-built for orchestrating AWS Glue jobs including crawlers and ETL tasks. They provide native integration with Glue components, built-in error handling, and are the most cost-effective option for this specific use case. Workflows eliminate the need for external orchestration tools and are optimized for Glue-based pipelines.
Why the other options are wrong
- A. AWS Step Functions is a general-purpose orchestration service with additional costs compared to Glue's native workflow capabilities.
- C. AWS Glue Studio is a visual design tool for creating jobs, not an orchestration service.
- D. Amazon MWAA is a managed Apache Airflow service that adds operational overhead and cost for a task that Glue workflows handle natively.
Question 7
A financial services company stores financial data in Amazon Redshift. A data engineer wants to run real-time queries on the financial data to support a web-based trading application. The data engineer wants to run the queries from within the trading application. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: B. Use the Amazon Redshift Data API.
The Amazon Redshift Data API allows applications to execute SQL queries without managing connections or database drivers, making it ideal for web-based applications. It requires minimal setup, handles authentication through IAM, and supports asynchronous query execution. This approach minimizes operational overhead compared to managing persistent connections or alternative data retrieval methods.
Why the other options are wrong
- A. WebSocket connections require custom implementation and management of persistent connections, adding operational complexity.
- C. JDBC connections require managing driver versions, connection pools, and network connectivity, adding operational overhead.
- D. Copying data to S3 and using S3 Select introduces latency and does not provide true real-time query capabilities on live Redshift data.
Question 8
A company uses Amazon Athena for one-time queries against data that is in Amazon S3. The company has several use cases. The company must implement permission controls to separate query processes and access to query history among users, teams, and applications that are in the same AWS account. Which solution will meet these requirements?
Show answer and explanation
Correct answer: B. Create an Athena workgroup for each use case. Apply tags to the workgroup. Create an IAM policy that uses the tags to apply appropriate permissions to the workgroup.
Athena workgroups provide separation of query processes, query history, and cost tracking per workgroup. By creating a workgroup for each use case and applying IAM policies that use tags to control permissions, the company can enforce fine-grained access control while separating concerns among users, teams, and applications within the same account. This is the most direct and purpose-built solution.
Why the other options are wrong
- A. Creating separate S3 buckets per use case is operationally complex and does not separate Athena query history or processes.
- C. IAM roles alone do not provide the workgroup-level separation of query processes and query history that Athena workgroups offer.
- D. AWS Glue Data Catalog resource policies control catalog access but do not provide Athena-specific controls for query history, processes, or cost separation.
Question 9
A data engineer needs to schedule a workflow that runs a set of AWS Glue jobs every day. The data engineer does not require the Glue jobs to run or finish at a specific time. Which solution will run the Glue jobs in the MOST cost-effective way?
Show answer and explanation
Correct answer: A. Choose the FLEX execution class in the Glue job properties.
The FLEX execution class is designed for non-time-sensitive workloads and uses spare capacity at a lower cost than STANDARD. Since the data engineer does not require the jobs to run at a specific time, FLEX execution class is the most cost-effective option. It provides significant cost savings for flexible, intermittent workloads by utilizing idle cluster resources.
Why the other options are wrong
- B. Spot Instances still incur compute costs and are not specifically designed for non-time-sensitive Glue jobs.
- C. STANDARD execution class is the default and more expensive option for time-sensitive workloads.
- D. Glue version selection does not directly impact cost-effectiveness for scheduling flexibility.
Question 10
A data engineer needs to create an AWS Lambda function that converts the format of data from .csv to Apache Parquet. The Lambda function must run only if a user uploads a .csv file to an Amazon S3 bucket. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: A. Create an S3 event notification that has an event type of s3:ObjectCreated:*. Use a filter rule to generate notifications only when the suffix includes .csv. Set the Amazon Resource Name (ARN) of the Lambda function as the destination for the event notification.
Option A provides the most direct and operationally efficient solution by using S3 event notifications with object creation events filtered by .csv suffix, directly invoking the Lambda function. This requires no intermediate services and automatically triggers the Lambda function whenever a .csv file is uploaded. It is the simplest architecture with minimal overhead.
Why the other options are wrong
- B. Using object tagging requires users to manually tag files, adding operational complexity and potential for human error.
- C. The s3:* event type is overly broad and would trigger on all S3 events, not just uploads, wasting resources.
- D. Adding SNS as an intermediary introduces unnecessary latency and operational overhead compared to direct Lambda invocation.
What people say about the full DEA-C01 pack 5.0 from 5
I passed the DEA-C01 exam with a score of 820/1000, and I couldn't have done it without CertStash! The dumps were spot-on with the exam content, and the explanations for each question were clear and helpful. Huge thanks to CertStash!
Just passed! The dump helped a lot.
Passed the exam! Tough exam but this dump helped a lot. Real questions in here. Went through it twice and passed comfortably.
Question 11
A data engineer needs to build an enterprise data catalog based on the company's Amazon S3 buckets and Amazon RDS databases. The data catalog must include storage format metadata for the data in the catalog. Which solution will meet these requirements with the LEAST effort?
Show answer and explanation
Correct answer: B. Use an AWS Glue crawler to build a data catalog. Use AWS Glue crawler classifiers to recognize the format of data and store the format in the catalog.
AWS Glue crawlers automatically scan S3 and RDS data sources to build a catalog. Glue crawler classifiers are specifically designed to recognize and classify data formats (CSV, JSON, Parquet, etc.) and automatically store format metadata in the catalog. This requires no manual intervention and directly solves the storage format metadata requirement with minimal effort.
Why the other options are wrong
- A. Requires manual inspection by data stewards, adding significant operational effort.
- C. Amazon Macie is designed for sensitive data detection, not data cataloging or format metadata collection.
- D. Requires custom script development and maintenance, which is more effort than using built-in Glue capabilities.
Question 12
A media company wants to use Amazon OpenSearch Service to analyze rea-time data about popular musical artists and songs. The company expects to ingest millions of new data events every day. The new data events will arrive through an Amazon Kinesis data stream. The company must transform the data and then ingest the data into the OpenSearch Service domain. Which method should the company use to ingest the data with the LEAST operational overhead?
Show answer and explanation
Correct answer: A. Use Amazon Kinesis Data Firehose and an AWS Lambda function to transform the data and deliver the transformed data to OpenSearch Service.
Amazon Kinesis Data Firehose with Lambda transformation provides the least operational overhead by handling the streaming ingestion, buffering, and delivery to OpenSearch Service automatically. Firehose manages scaling, retries, and error handling without requiring infrastructure provisioning. Lambda functions provide flexible transformation logic while Firehose manages all the operational complexity of data delivery.
Why the other options are wrong
- B. Logstash pipelines require provisioning and managing additional infrastructure outside of AWS managed services.
- C. Kinesis Agent is a client-side tool that adds operational complexity; Lambda would need to manage both agent invocations and delivery logic.
- D. KCL requires writing custom consumer code and managing consumer application infrastructure, adding significant operational overhead.
Question 13
A company needs to send customer call data from its on-premises PostgreSQL database to AWS to generate near real-time insights. The solution must capture and load updates from operational data stores that run in the PostgreSQL database. The data changes continuously. A data engineer configures an AWS Database Migration Service (AWS DMS) ongoing replication task. The task reads changes in near real time from the PostgreSQL source database transaction logs for each table. The task then sends the data to an Amazon Redshift cluster for processing. The data engineer discovers latency issues during the change data capture (CDC) of the task. The data engineer thinks that the PostgreSQL source database is causing the high latency. Which solution will confirm that the PostgreSQL database is the source of the high latency?
Show answer and explanation
Correct answer: D. Use Amazon CloudWatch to monitor the DMS task. Examine the CDCLatencySource metric to identify delays in the CDC from the source database.
The CDCLatencySource metric in CloudWatch specifically measures the latency introduced by the source database during change data capture. This metric directly identifies delays in CDC from the source PostgreSQL database, making it the most appropriate diagnostic tool for confirming if the database is causing high latency.
Why the other options are wrong
- A. CDCIncomingChanges metric tracks the volume of incoming changes, not the latency of the source database operations.
- B. Verifying logical replication configuration is a setup check, not a latency diagnostic tool.
- C. CloudWatch Logs errors indicate configuration or failure issues, not latency problems from the source database.
Question 14
A data engineer needs to maintain a central metadata repository that users access through Amazon EMR and Amazon Athena queries. The repository needs to provide the schema and properties of many tables. Some of the metadata is stored in Apache Hive. The data engineer needs to import the metadata from Hive into the central metadata repository. Which solution will meet these requirements with the LEAST development effort?
Show answer and explanation
Correct answer: C. Use the AWS Glue Data Catalog.
AWS Glue Data Catalog is the purpose-built AWS service for managing metadata centrally and is fully integrated with both Amazon EMR and Amazon Athena. It can import Hive metastore metadata directly and provides a unified metadata repository with minimal development effort. Users can query through Athena and EMR without additional configuration.
Why the other options are wrong
- A. Apache Ranger is for access control, not metadata repository management.
- B. A Hive metastore on EMR is distributed and not centralized, defeating the requirement for a central repository.
- D. RDS for MySQL requires custom schema design and development effort to build a metadata repository, lacking native integration with Athena and EMR.
Question 15
A retail company is expanding its operations globally. The company needs to use Amazon QuickSight to accurately calculate currency exchange rates for financial reports. The company has an existing dashboard that includes a visual that is based on an analysis of a dataset that contains global currency values and exchange rates. A data engineer needs to ensure that exchange rates are calculated with a precision of four decimal places. The calculations must be precomputed. The data engineer must materialize results in QuickSight super-fast, parallel, in-memory calculation engine (SPICE). Which solution will meet these requirements?
Show answer and explanation
Correct answer: A. Define and create the calculated field in the dataset.
Calculated fields defined at the dataset level are precomputed and materialized in SPICE, enabling fast parallel in-memory calculations with consistent precision across all analyses and visuals that use the dataset. This ensures exchange rates are calculated once with four decimal place precision and reused efficiently across the dashboard.
Why the other options are wrong
- B. Analysis-level calculated fields are computed per analysis and not materialized in SPICE for reuse across other analyses.
- C. Visual-level calculated fields are computed at query time per visual and do not benefit from SPICE precomputation and materialization.
- D. Dashboard-level calculated fields do not exist; calculations must be defined at dataset, analysis, or visual levels.
Question 16
A company stores daily records of the financial performance of investment portfolios in .csv format in an Amazon S3 bucket. A data engineer uses AWS Glue crawlers to crawl the S3 data. The data engineer must make the S3 data accessible daily in the AWS Glue Data Catalog. Which solution will meet these requirements?
Show answer and explanation
Correct answer: B. Create an IAM role that includes the AWSGlueServiceRole policy. Associate the role with the crawler. Specify the S3 bucket path of the source data as the crawler's data store. Create a daily schedule to run the crawler. Specify a database name for the output.
The correct configuration requires the AWSGlueServiceRole (which includes necessary S3 and Glue permissions), scheduling the crawler to run daily, and specifying a database name for output to the Data Catalog. The crawler stores metadata in the Data Catalog database, not in S3. AWSGlueServiceRole is the proper managed role with appropriate permissions for Glue operations.
Why the other options are wrong
- A. AmazonS3FullAccess is overly permissive and not the recommended role; also outputs to S3 path instead of Data Catalog database.
- C. AmazonS3FullAccess is not the proper Glue role; DPUs are for Glue job execution, not crawler scheduling.
- D. Outputs to S3 path instead of Data Catalog database, which does not make data accessible through the Data Catalog as required.
Question 17
A company plans to use Amazon Kinesis Data Firehose to store data in Amazon S3. The source data consists of 2 MB .csv files. The company must convert the .csv files to JSON format. The company must store the files in Apache Parquet format. Which solution will meet these requirements with the LEAST development effort?
Show answer and explanation
Correct answer: D. Use Kinesis Data Firehose to invoke an AWS Lambda function that transforms the .csv files to JSON. Use Kinesis Data Firehose to store the files in Parquet format.
Kinesis Data Firehose record format conversion can convert JSON input into Apache Parquet, but it cannot read .csv input directly. A Firehose data transformation Lambda function converts the .csv records to JSON, and the built-in format conversion then writes Parquet with no further code.
Why the other options are wrong
- A. Firehose cannot convert .csv to JSON natively, and writing Parquet in Lambda duplicates a feature Firehose already provides.
- B. Firehose format conversion only accepts JSON input, so it cannot convert the .csv files without a transformation function.
- C. Writing Parquet inside the Lambda function adds custom development that the built-in Firehose format conversion already handles.
Question 18
A company implements a data mesh that has a central governance account. The company needs to catalog all data in the governance account. The governance account uses AWS Lake Formation to centrally share data and grant access permissions. The company has created a new data product that includes a group of Amazon Redshift Serverless tables. A data engineer needs to share the data product with a marketing team. The marketing team must have access to only a subset of columns. The data engineer needs to share the same data product with a compliance team. The compliance team must have access to a different subset of columns than the marketing team needs access to. Which combination of steps should the data engineer take to meet these requirements? (Choose two.)
Show answer and explanation
Correct answer: B, D. Create an Amazon Redshift data share that includes the tables that need to be shared; Share the Amazon Redshift data share to the Lake Formation catalog in the governance account.
Option B is essential: create an Amazon Redshift data share to enable sharing capabilities with external accounts. Option D completes the solution by sharing the data share to the Lake Formation catalog, which allows governance and cross-account sharing from the central governance account. Options A and C address views and VPC endpoints but views alone don't integrate with Lake Formation's centralized governance. Option E shares to workgroups rather than through the governance mechanism required by Lake Formation.
Why the other options are wrong
- A. Views are helpful for column-level access but must be combined with proper data share mechanisms for Lake Formation integration.
- C. While VPC endpoints enable connectivity, they don't provide the Lake Formation governance integration required for the centralized data mesh setup.
- E. Sharing directly to workgroups bypasses the Lake Formation governance catalog required by the centralized governance account architecture.
Question 19
An ecommerce company operates a complex order fulfilment process that spans several operational systems hosted in AWS. Each of the operational systems has a Java Database Connectivity (JDBC)-compliant relational database where the latest processing state is captured. The company needs to give an operations team the ability to track orders on an hourly basis across the entire fulfillment process. Which solution will meet these requirements with the LEAST development overhead?
Show answer and explanation
Correct answer: A. Use AWS Glue to build ingestion pipelines from the operational systems into Amazon Redshift Build dashboards in Amazon QuickSight that track the orders.
AWS Glue with Amazon Redshift provides the least development overhead for this analytical use case. Glue easily handles JDBC connections to multiple operational databases, provides native ETL capabilities, and Redshift is optimized for complex analytical queries across multiple systems with built-in columnar storage and aggregation. QuickSight integrates seamlessly for dashboard development. DynamoDB (option B) is not ideal for complex analytical aggregations across multiple systems. Options C and D use DMS for change capture and cross-region replication, adding unnecessary operational complexity; DMS is better suited for real-time synchronization than batch hourly analytics, and Grafana (option C) requires additional setup compared to native AWS QuickSight.
Why the other options are wrong
- B. DynamoDB is not optimized for complex analytical queries and aggregations needed for hourly order tracking across systems.
- C. DMS with cross-region replication adds complexity; using Grafana requires additional setup outside AWS native tools.
- D. DMS is overkill for hourly batch analytics and introduces unnecessary operational overhead with cross-region replication setup.
Question 20
A company maintains multiple extract, transform, and load (ETL) workflows that ingest data from the company's operational databases into an Amazon S3 based data lake. The ETL workflows use AWS Glue and Amazon EMR to process data. The company wants to improve the existing architecture to provide automated orchestration and to require minimal manual effort. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: B. AWS Step Functions tasks
AWS Step Functions provides a serverless, declarative orchestration service specifically designed for complex workflows integrating multiple AWS services like Glue and EMR. Step Functions handles state management, error handling, retries, and parallelization automatically with minimal operational overhead. It integrates natively with both Glue and EMR and requires no infrastructure management, making it the least operationally burdensome solution for automated orchestration.
Why the other options are wrong
- A. Glue workflows are basic orchestration; Step Functions provides more advanced features and better integration across Glue and EMR.
- C. Lambda requires manual orchestration and error handling logic, introducing significant operational complexity.
- D. MWAA adds overhead through Airflow management; Step Functions is simpler and more AWS-native for this use case.
Question 21
A company uses Amazon RDS for MySQL as the database for a critical application. The database workload is mostly writes, with a small number of reads. A data engineer notices that the CPU utilization of the DB instance is very high. The high CPU utilization is slowing down the application. The data engineer must reduce the CPU utilization of the DB Instance. Which actions should the data engineer take to meet this requirement? (Choose two.)
Show answer and explanation
Correct answer: A, D. Use the Performance Insights feature of Amazon RDS to identify queries that have high CPU utilization. Optimize the problematic queries; Upgrade to a larger instance size.
Using Performance Insights (option A) identifies the specific high-CPU queries causing the bottleneck, allowing targeted optimization of problematic queries for maximum impact. Upgrading to a larger instance size (option D) provides immediate relief by distributing the write-heavy workload across more CPU resources. Both actions directly address the CPU utilization problem.
Why the other options are wrong
- B. Adding more tables and indexes typically increases CPU usage for writes and maintenance, worsening the situation.
- C. Weekly reboots do not address the underlying cause of high CPU utilization and provide no lasting benefit.
- E. Caching reduces reads but the workload is already mostly writes, so caching has minimal impact on write-heavy CPU usage.
Question 22
A company is using an AWS Transfer Family server to migrate data from an on-premises environment to AWS. Company policy mandates the use of TLS 1.2 or above to encrypt the data in transit. Which solution will meet these requirements?
Show answer and explanation
Correct answer: C. Update the security policy of the Transfer Family server to specify a minimum protocol version of TLS 1.2
AWS Transfer Family servers have a security policy configuration that allows you to specify the minimum protocol version for TLS encryption. Setting the security policy to require TLS 1.2 or above is the direct and proper way to enforce this company policy requirement. This is a built-in feature of the Transfer Family service specifically designed for this purpose.
Why the other options are wrong
- A. SSH keys are for key-based authentication, not for enforcing TLS protocol versions
- B. Security group rules control network access, not TLS protocol version enforcement
- D. SSL certificates enable encryption but don't enforce a minimum TLS version; the security policy is what enforces protocol requirements
Question 23
A company maintains a data warehouse in an on-premises Oracle database. The company wants to build a data lake on AWS. The company wants to load data warehouse tables into Amazon S3 and synchronize the tables with incremental data that arrives from the data warehouse every day. Each table has a column that contains monotonically increasing values. The size of each table is less than 50 GB. The data warehouse tables are refreshed every night between 1 AM and 2 AM. A business intelligence team queries the tables between 10 AM and 8 PM every day. Which solution will meet these requirements in the MOST operationally efficient way?
Show answer and explanation
Correct answer: A. Use an AWS Database Migration Service (AWS DMS) full load plus CDC job to load tables that contain monotonically increasing data columns from the on-premises data warehouse to Amazon S3. Use custom logic in AWS Glue to append the daily incremental data to a fullload copy that is in Amazon S3.
AWS DMS with full load plus CDC (Change Data Capture) is purpose-built for this scenario: it performs an initial full load to S3, then continuously captures and syncs incremental changes from the Oracle database using the monotonically increasing column as the change indicator. Custom Glue logic then appends the daily CDC data. This approach minimizes data transfer (only changes are captured), aligns with the nightly refresh window, and supports the daytime query window without conflicts. It's more operationally efficient than daily full reloads and leverages AWS's migration tool designed for this exact use case.
Why the other options are wrong
- B. While Glue JDBC with bookmarks could work, it requires more custom logic and may not capture all incremental changes as reliably as DMS CDC.
- C. Daily full-load overwrite is operationally inefficient for 50 GB tables and wastes bandwidth by reloading unchanged data.
- D. Glue full-load daily overwrite similarly wastes resources and doesn't leverage change detection for incremental synchronization.
Question 24
A data engineer configured an AWS Glue Data Catalog for data that is stored in Amazon S3 buckets. The data engineer needs to configure the Data Catalog to receive incremental updates. The data engineer sets up event notifications for the S3 bucket and creates an Amazon Simple Queue Service (Amazon SQS) queue to receive the S3 events. Which combination of steps should the data engineer take to meet these requirements with LEAST operational overhead? (Choose two.)
Show answer and explanation
Correct answer: A, C. Create an S3 event-based AWS Glue crawler to consume events from the SQS queue; Use an AWS Lambda function to directly update the Data Catalog based on S3 events that the SQS queue receives.
An S3 event-based AWS Glue crawler (A) consumes SQS events triggered by S3 uploads and automatically updates the Data Catalog, eliminating manual triggers and schedule maintenance. Using an AWS Lambda function (C) to directly update the Data Catalog based on SQS events provides an alternative event-driven approach. Both achieve incremental updates with minimal operational overhead compared to scheduling or manual approaches.
Why the other options are wrong
- B. Time-based schedules require ongoing management and do not respond to actual data changes in real-time.
- D. Manual initiation requires constant monitoring and human intervention, creating operational overhead.
- E. Step Functions adds orchestration complexity without providing operational benefits over event-based crawlers or Lambda functions.
Question 25
A company stores petabytes of data in thousands of Amazon S3 buckets in the S3 Standard storage class. The data supports analytics workloads that have unpredictable and variable data access patterns. The company does not access some data for months. However, the company must be able to retrieve all data within milliseconds. The company needs to optimize S3 storage costs. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: D. Use S3 Intelligent-Tiering. Use the default access tier.
S3 Intelligent-Tiering automatically transitions objects between access tiers based on actual usage patterns without manual intervention. With unpredictable and variable access patterns, it intelligently moves frequently accessed data to cheaper tiers and infrequently accessed data to deeper archives. The default access tier includes frequent, infrequent, and archive tiers, providing automatic cost optimization. No operational overhead is required for monitoring or maintaining lifecycle policies. Options A and B require continuous manual refinement; option C activates Deep Archive which violates the millisecond retrieval requirement.
Why the other options are wrong
- A. Requires ongoing manual analysis and policy refinement, creating continuous operational overhead.
- B. Requires manual policy configuration and ongoing refinement to optimize costs based on variable access patterns.
- C. Deep Archive has retrieval times of 12+ hours, violating the millisecond retrieval requirement.
Question 26
A company receives test results from testing facilities that are located around the world. The company stores the test results in millions of 1 KB JSON files in an Amazon S3 bucket. A data engineer needs to process the files, convert them into Apache Parquet format, and load them into Amazon Redshift tables. The data engineer uses AWS Glue to process the files, AWS Step Functions to orchestrate the processes, and Amazon EventBridge to schedule jobs. The company recently added more testing facilities. The time required to process files is increasing. The data engineer must reduce the data processing time. Which solution will MOST reduce the data processing time?
Show answer and explanation
Correct answer: B. Use the AWS Glue dynamic frame file-grouping option to ingest the raw input files. Process the files. Load the files into the Amazon Redshift tables.
The bottleneck with millions of 1 KB files is the overhead of processing small files individually. AWS Glue's dynamic frame file-grouping option is specifically designed to group small input files into larger partitions before processing, dramatically reducing the number of tasks and I/O operations. This is the most operationally efficient solution as it leverages native Glue functionality without requiring additional services or custom logic.
Why the other options are wrong
- A. While Lambda grouping would work, it adds operational complexity and requires managing intermediate storage; Glue's native feature is simpler.
- C. Loading raw JSON directly into Redshift via COPY without transformation defeats the purpose of converting to Parquet and lacks data validation.
- D. EMR would require additional infrastructure management and wouldn't inherently solve the small-file problem better than Glue's built-in grouping.
Question 27
A car sales company maintains data about cars that are listed for sale in an area. The company receives data about new car listings from vendors who upload the data daily as compressed files into Amazon S3. The compressed files are up to 5 KB in size. The company wants to see the most up-to-date listings as soon as the data is uploaded to Amazon S3. A data engineer must automate and orchestrate the data processing workflow of the listings to feed a dashboard. The data engineer must also provide the ability to perform one-time queries and analytical reporting. The query solution must be scalable. Which solution will meet these requirements MOST cost-effectively?
Show answer and explanation
Correct answer: D. Use AWS Glue to process incoming data. Use AWS Lambda and S3 Event Notifications to orchestrate workflows. Use Amazon Athena for one-time queries and analytical reporting. Use Amazon QuickSight for the dashboard.
AWS Glue serverlessly processes incoming compressed files with no cluster management. S3 Event Notifications and Lambda provide automatic, event-driven orchestration without Step Functions overhead. Amazon Athena provides scalable, on-demand querying at low cost without provisioning. Amazon QuickSight creates the real-time dashboard. This is the most cost-effective because it uses serverless components with pay-per-query/use pricing and minimal operational overhead.
Why the other options are wrong
- A. EMR and OpenSearch require provisioned resources with significant operational overhead and higher costs than serverless alternatives.
- B. Provisioned EMR cluster incurs continuous costs even when idle; Step Functions adds unnecessary orchestration complexity over event-driven approaches.
- C. Redshift Spectrum is more expensive than Athena for ad-hoc queries; OpenSearch requires provisioned instances with ongoing management.
Question 28
A company is building a data lake for a new analytics team. The company is using Amazon S3 for storage and Amazon Athena for query analysis. All data that is in Amazon S3 is in Apache Parquet format. The company is running a new Oracle database as a source system in the company’s data center. The company has 70 tables in the Oracle database. All the tables have primary keys. Data can occasionally change in the source system. The company wants to ingest the tables every day into the data lake. Which solution will meet this requirement with the LEAST effort?
Show answer and explanation
Correct answer: C. Create an AWS Database Migration Service (AWS DMS) task for ongoing replication. Set the Oracle database as the source. Set Amazon S3 as the target. Configure the task to write the data in Parquet format.
AWS DMS can use an on-premises Oracle database as the source and Amazon S3 as the target, writing Parquet files. An ongoing replication task uses change data capture, so inserts and updates to all 70 tables flow to the data lake with little setup and no custom code.
Why the other options are wrong
- A. Sqoop on Amazon EMR requires running a cluster and building jobs, and it does not capture changed rows easily.
- B. Glue job bookmarks track new data by key or timestamp and do not capture updates to existing rows, so changed records would be missed.
- D. Migrating to RDS and building triggers that call Lambda functions adds a new database and custom code, which is far more effort.
Question 29
A company is creating near real-time dashboards to visualize time series data. The company ingests data into Amazon Managed Streaming for Apache Kafka (Amazon MSK). A customized data pipeline consumes the data. The pipeline then writes data to Amazon Keyspaces (for Apache Cassandra), Amazon OpenSearch Service, and Apache Avro objects in Amazon S3. Which solution will make the data available for the data visualizations with the LEAST latency?
Show answer and explanation
Correct answer: A. Create OpenSearch Dashboards by using the data from OpenSearch Service.
OpenSearch Dashboards accessing data directly from OpenSearch Service provides the lowest latency for near real-time visualization. Since data is already being written to OpenSearch Service in the pipeline, querying it directly eliminates additional hops and transformations. Athena and S3 queries introduce additional latency by requiring query execution time, while OpenSearch Dashboards can leverage the already-indexed data in OpenSearch Service for immediate visualization.
Why the other options are wrong
- B. Using Athena with S3 Avro objects adds query latency compared to direct OpenSearch access and requires Hive metastore overhead.
- C. Querying S3 through Athena introduces query execution latency and complexity with Keyspaces as a catalog, which is not its intended purpose.
- D. S3 Select and Glue add processing steps and latency compared to querying pre-indexed data already in OpenSearch Service.
Question 30
A data engineer needs to use an Amazon QuickSight dashboard that is based on Amazon Athena queries on data that is stored in an Amazon S3 bucket. When the data engineer connects to the QuickSight dashboard, the data engineer receives an error message that indicates insufficient permissions. Which factors could cause to the permissions-related errors? (Choose two.)
Show answer and explanation
Correct answer: C, D. QuickSight does not have access to the S3 bucket; QuickSight does not have access to decrypt S3 data.
QuickSight requires two key permissions to access Athena-based data: it must have access to the underlying S3 bucket where the data resides (option C), and it must have permission to decrypt the S3 data if it is encrypted (option D). Without these permissions, QuickSight cannot successfully query the data through Athena.
Why the other options are wrong
- A. QuickSight connects to Athena through the Athena API and metadata, not through a direct 'connection' configuration that would cause permission errors.
- B. Athena tables derive their permissions from S3 access and IAM policies; uncataloged tables would cause query failures, not permission errors in QuickSight.
- E. QuickSight uses IAM roles for authentication, but the permission error would stem from specific S3 and encryption access, not the role's absence.
Question 31
A company has used an Amazon Redshift table that is named Orders for 6 months. The company performs weekly updates and deletes on the table. The table has an interleaved sort key on a column that contains AWS Regions. The company wants to reclaim disk space so that the company will not run out of storage space. The company also wants to analyze the sort key column. Which Amazon Redshift command will meet these requirements?
Show answer and explanation
Correct answer: C. VACUUM REINDEX Orders
VACUUM REINDEX is the appropriate command for reclaiming disk space and re-sorting an interleaved sort key. It performs both the space reclamation that regular VACUUM provides and reindexes the interleaved sort key column, which optimizes for future queries on that Region column. This meets both requirements: freeing storage and analyzing the sort key.
Why the other options are wrong
- A. VACUUM FULL reclaims space but does not reindex interleaved sort keys; it's also slower than other options.
- B. VACUUM DELETE ONLY only reclaims space from deleted rows but does not address sort key analysis or optimization.
- D. VACUUM SORT ONLY is not a valid Redshift command; the correct command for sort key optimization is VACUUM REINDEX.
Question 32
A company has AWS resources in multiple AWS Regions. The company has an Amazon EFS file system in each Region where the company operates. The company’s data science team operates within only a single Region. The data that the data science team works with must remain within the team's Region. A data engineer needs to create a single dataset by processing files that are in each of the company's Regional EFS file systems. The data engineer wants to use an AWS Step Functions state machine to orchestrate AWS Lambda functions to process the data. Which solution will meet these requirements with the LEAST effort?
Show answer and explanation
Correct answer: C. Deploy the Lambda functions to each Region. Mount the Regional EFS file systems to the Lambda functions. Use the Lambda functions to process the data. Store the output in an Amazon S3 bucket in the data science team’s Region.
Option C deploys Lambda functions to each Region to process data locally from their respective EFS file systems, then stores output in a central S3 bucket in the data science team's Region. This satisfies the requirement that data must remain within each Region during processing while still allowing a single dataset to be created. Options A and B violate the data residency requirement by bringing data into the data science team's Region. Option D requires data transfer which is less efficient than processing data where it resides.
Why the other options are wrong
- A. VPC peering and cross-Region EFS mounting would violate the requirement that data must remain within the data science team's Region.
- B. Replicating EFS data to the data science team's Region violates the data residency requirement.
- D. DataSync transfer adds unnecessary complexity and data movement compared to processing data in-place with distributed Lambda functions.
Question 33
A financial company recently added more features to its mobile app. The new features required the company to create a new topic in an existing Amazon Managed Streaming for Apache Kafka (Amazon MSK) cluster. A few days after the company added the new topic, Amazon CloudWatch raised an alarm on the RootDiskUsed metric for the MSK cluster. How should the company address the CloudWatch alarm?
Show answer and explanation
Correct answer: A. Expand the storage of the MSK broker. Configure the MSK cluster storage to expand automatically.
The RootDiskUsed alarm indicates insufficient broker disk space, which is needed for storing Kafka topic data and logs. Expanding MSK broker storage and enabling automatic expansion prevents future disk space issues when new topics are added or data volume increases, directly addressing the root cause of the alarm.
Why the other options are wrong
- B. ZooKeeper nodes store cluster metadata, not topic data; disk issues on brokers are unrelated to ZooKeeper storage.
- C. Changing instance types does not expand storage capacity and requires cluster restart, while the issue is specifically disk space exhaustion.
- D. The Target Volume-in-GiB parameter does not exist for MSK topics; storage is managed at the broker level, not per topic.
Question 34
A data engineer uses Amazon Redshift to run resource-intensive analytics processes once every month. Every month, the data engineer creates a new Redshift provisioned cluster. The data engineer deletes the Redshift provisioned cluster after the analytics processes are complete every month. Before the data engineer deletes the cluster each month, the data engineer unloads backup data from the cluster to an Amazon S3 bucket. The data engineer needs a solution to run the monthly analytics processes that does not require the data engineer to manage the infrastructure manually. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: B. Use Amazon Redshift Serverless to automatically process the analytics workload.
Amazon Redshift Serverless automatically manages infrastructure provisioning, scaling, and de-provisioning based on workload demands. It eliminates the need to manually create and delete clusters monthly, making it ideal for sporadic, resource-intensive workloads. The service handles capacity management automatically, requiring zero manual infrastructure management. Options A, C, and D all require manual cluster lifecycle management or custom scripting; Serverless eliminates this entirely.
Why the other options are wrong
- A. Pausing and resuming still requires manual cluster creation and deletion; does not automate the full lifecycle.
- C. AWS CLI requires writing and maintaining custom scripts for the full cluster lifecycle.
- D. CloudFormation requires creating and updating templates manually for each monthly run; does not eliminate operational overhead.
Question 35
A company has five offices in different AWS Regions. Each office has its own human resources (HR) department that uses a unique IAM role. The company stores employee records in a data lake that is based on Amazon S3 storage. A data engineering team needs to limit access to the records. Each HR department should be able to access records for only employees who are within the HR department's Region. Which combination of steps should the data engineering team take to meet this requirement with the LEAST operational overhead? (Choose two.)
Show answer and explanation
Correct answer: B, D. Register the S3 path as an AWS Lake Formation location; Enable fine-grained access control in AWS Lake Formation. Add a data filter for each Region.
AWS Lake Formation with fine-grained access control (option D) provides the least operational overhead by enabling centralized permission management. Registering the S3 path as a Lake Formation location (option B) is necessary to enable Lake Formation to manage permissions. This combination allows defining data filters by Region within Lake Formation, automatically enforcing access restrictions for each HR department's IAM role without modifying individual IAM policies or creating separate buckets.
Why the other options are wrong
- A. Data filters must be registered as Lake Formation locations, not just used as generic S3 paths.
- C. Modifying individual IAM roles increases operational overhead and is not the least administrative approach.
- E. Creating separate S3 buckets per region increases operational complexity and management overhead significantly.
Question 36
An airline company is collecting metrics about flight activities for analytics. The company is conducting a proof of concept (POC) test to show how analytics can provide insights that the company can use to increase on-time departures. The POC test uses objects in Amazon S3 that contain the metrics in .csv format. The POC test uses Amazon Athena to query the data. The data is partitioned in the S3 bucket by date. As the amount of data increases, the company wants to optimize the storage solution to improve query performance. Which combination of solutions will meet these requirements? (Choose two.)
Show answer and explanation
Correct answer: C, E. Use an S3 bucket that is in the same AWS Region where the company runs Athena queries; Preprocess the .csv data to Apache Parquet format by fetching only the data blocks that are needed for predicates.
Converting CSV data to Apache Parquet format (option E) dramatically improves query performance by enabling columnar storage and predicate pushdown, allowing Athena to read only required data blocks. Ensuring the S3 bucket is in the same AWS Region as Athena queries (option C) eliminates cross-region latency. Both optimizations directly address query performance without significant cost or complexity.
Why the other options are wrong
- A. Randomizing key prefixes is beneficial for write throughput but does not improve query performance for read-heavy analytical workloads.
- B. The S3 bucket account matters less than region for query performance; the key issue is latency, not account boundaries.
- D. Preprocessing to JSON format does not improve query performance; it actually increases data size compared to Parquet and does not enable predicate pushdown like Parquet does.
Question 37
A data engineer needs to use AWS Step Functions to design an orchestration workflow. The workflow must parallel process a large collection of data files and apply a specific transformation to each file. Which Step Functions state should the data engineer use to meet these requirements?
Show answer and explanation
Correct answer: C. Map state
The Map state in AWS Step Functions is specifically designed to iterate over a collection and run steps in parallel for each item. It's the native solution for distributing work across multiple items concurrently, making it perfect for applying transformations to a large collection of data files in parallel.
Why the other options are wrong
- A. Parallel state runs different branches concurrently but is not designed for iterating over collections of items.
- B. Choice state makes conditional decisions about workflow paths; it doesn't perform parallel processing.
- D. Wait state pauses execution for a specified duration; it has no role in parallel data processing.
Question 38
An online retail company has an application that runs on Amazon EC2 instances that are in a VPC. The company wants to collect flow logs for the VPC and analyze network traffic. Which solution will meet these requirements MOST cost-effectively?
Show answer and explanation
Correct answer: D. Publish flow logs to Amazon S3 in Apache Parquet format. Use Amazon Athena for analytics.
Publishing flow logs to S3 in Apache Parquet format and using Athena is most cost-effective because Parquet is a columnar format that compresses significantly better than text, reducing S3 storage costs. Athena charges by data scanned, so the compression savings translate directly to lower query costs compared to text format or CloudWatch Logs storage fees.
Why the other options are wrong
- A. CloudWatch Logs storage has per-GB ingestion and retention costs that accumulate quickly with high-volume flow logs.
- B. OpenSearch Service clusters require provisioned infrastructure costs, making them more expensive than serverless Athena for simple analytics.
- C. Text format in S3 does not compress as well as Parquet, resulting in higher storage and query costs in Athena.
Question 39
A company stores logs in an Amazon S3 bucket. When a data engineer attempts to access several log files, the data engineer discovers that some files have been unintentionally deleted. The data engineer needs a solution that will prevent unintentional file deletion in the future. Which solution will meet this requirement with the LEAST operational overhead?
Show answer and explanation
Correct answer: B. Enable S3 Versioning for the S3 bucket.
S3 Versioning is the LEAST operationally overhead solution because it is a simple toggle that automatically maintains multiple versions of objects. When enabled, deleted files are not permanently removed but marked as deleted, allowing recovery of previous versions. This requires no additional backup infrastructure, scheduling, or management compared to other options.
Why the other options are wrong
- A. Manual backups require ongoing operational effort and scheduling discipline
- C. Replication requires configuration and monitoring of replica buckets, adding operational complexity
- D. Glacier archival is for long-term retention of historical data, not for preventing unintentional deletion of active files
Question 40
Files from multiple data sources arrive in an Amazon S3 bucket on a regular basis. A data engineer wants to ingest new files into Amazon Redshift in near real time when the new files arrive in the S3 bucket. Which solution will meet these requirements?
Show answer and explanation
Correct answer: D. Use S3 Event Notifications to invoke an AWS Lambda function that loads new files into Amazon Redshift.
S3 Event Notifications automatically trigger when new files arrive in the S3 bucket, invoking a Lambda function that immediately processes and loads the files into Redshift. This provides true near-real-time ingestion with minimal latency. The query editor scheduling requires manual configuration and is not event-driven; Aurora zero-ETL is for Aurora-to-Redshift integration not S3 files; Glue job bookmarks require scheduled execution rather than event-driven triggers.
Why the other options are wrong
- A. Scheduled COPY commands have fixed intervals and cannot respond immediately to file arrivals.
- B. Aurora zero-ETL integrates Aurora databases with Redshift, not S3 file ingestion.
- C. Glue job bookmarks work with scheduled jobs, not real-time event-driven ingestion.
Question 41
A company's data engineer needs to optimize the performance of table SQL queries. The company stores data in an Amazon Redshift cluster. The data engineer cannot increase the size of the cluster because of budget constraints. The company stores the data in multiple tables and loads the data by using the EVEN distribution style. Some tables are hundreds of gigabytes in size. Other tables are less than 10 MB in size. Which solution will meet these requirements?
Show answer and explanation
Correct answer: C. Use the ALL distribution style for rarely updated small tables. Specify primary and foreign keys for all tables.
The ALL distribution style copies entire small tables to every node, eliminating join shuffles and improving query performance for small dimension tables. Using ALL for the rarely-updated small tables (less than 10 MB) provides this benefit without significant overhead, while keeping EVEN for large tables preserves cluster balance. Specifying primary and foreign keys enables Redshift's query optimizer to use this metadata. Option A ignores the performance problem; option B wastes resources applying ALL to massive tables; option D overly complicates the design without clear benefit for this scenario.
Why the other options are wrong
- A. EVEN distribution for small tables causes unnecessary join shuffles and poor query performance.
- B. ALL distribution for hundreds-of-gigabyte tables wastes storage and network resources across nodes.
- D. Partition keys are not applicable to Redshift tables, and partition/sort keys are not the primary solution for this distribution problem.
Question 42
A company is building an analytics solution. The solution uses Amazon S3 for data lake storage and Amazon Redshift for a data warehouse. The company wants to use Amazon Redshift Spectrum to query the data that is in Amazon S3. Which actions will provide the FASTEST queries? (Choose two.)
Show answer and explanation
Correct answer: B, C. Use a columnar storage file format; Partition the data based on the most common query predicates.
Columnar storage formats like Parquet and ORC allow Spectrum to read only relevant columns, significantly reducing data scanned. Partitioning data by common query predicates enables partition pruning, allowing Spectrum to skip entire file partitions that don't match filter conditions. Both techniques reduce I/O and dramatically speed up query execution.
Why the other options are wrong
- A. Gzip compression between 1-5 GB is reasonable but doesn't directly optimize query speed like columnar formats or partitioning do.
- D. Files smaller than 10 KB create excessive overhead; larger files are more efficient for parallel processing.
- E. Non-splittable formats prevent parallel processing of files, reducing query performance instead of improving it.
Question 43
A company uploads .csv files to an Amazon S3 bucket. The company’s data platform team has set up an AWS Glue crawler to perform data discovery and to create the tables and schemas. An AWS Glue job writes processed data from the tables to an Amazon Redshift database. The AWS Glue job handles column mapping and creates the Amazon Redshift tables in the Redshift database appropriately. If the company reruns the AWS Glue job for any reason, duplicate records are introduced into the Amazon Redshift tables. The company needs a solution that will update the Redshift tables without duplicates. Which solution will meet these requirements?
Show answer and explanation
Correct answer: A. Modify the AWS Glue job to copy the rows into a staging Redshift table. Add SQL commands to update the existing rows with new values from the staging Redshift table.
The recommended approach to handle reruns without duplicates is to use a staging table pattern. The AWS Glue job loads data into a staging Redshift table, then SQL commands are executed to update existing rows with new values from the staging table (using UPSERT-like operations or DELETE-INSERT patterns). This ensures idempotency and prevents duplicate records when the job is rerun.
Why the other options are wrong
- B. Using an external MySQL database as an intermediate step adds unnecessary complexity and latency; Redshift can perform upsert operations directly.
- C. The dropDuplicates() API only removes duplicates within the current dataset but does not prevent duplicates from being inserted into Redshift on reruns.
- D. ResolveChoice is a transform for handling schema conflicts during data discovery, not for preventing duplicates during data loading.
Question 44
A company stores data from an application in an Amazon DynamoDB table that operates in provisioned capacity mode. The workloads of the application have predictable throughput load on a regular schedule. Every Monday, there is an immediate increase in activity early in the morning. The application has very low usage during weekends. The company must ensure that the application performs consistently during peak usage times. Which solution will meet these requirements in the MOST cost-effective way?
Show answer and explanation
Correct answer: C. Use AWS Application Auto Scaling to schedule higher provisioned capacity for peak usage times. Schedule lower capacity during off-peak times.
AWS Application Auto Scaling supports scheduled scaling for DynamoDB, allowing the company to automatically provision higher capacity on Monday mornings and lower capacity during weekends. This matches the predictable schedule exactly, ensuring consistent performance during peaks while minimizing costs during low-usage periods, more cost-effective than maintaining peak capacity constantly.
Why the other options are wrong
- A. Provisioning to maximum peak capacity all the time wastes money on unused capacity during low-usage periods like weekends.
- B. Splitting tables adds complexity without solving the core scaling issue and doesn't optimize for predictable scheduled patterns.
- D. On-demand mode is less cost-effective than scheduled provisioning for predictable, recurring workload patterns with known peak times.
Question 45
A company has a data lake in Amazon S3. The company uses AWS Glue to catalog data and AWS Glue Studio to implement data extract, transform, and load (ETL) pipelines. The company needs to ensure that data quality issues are checked every time the pipelines run. A data engineer must enhance the existing pipelines to evaluate data quality rules based on predefined thresholds. Which solution will meet these requirements with the LEAST implementation effort?
Show answer and explanation
Correct answer: B. Add a new Evaluate Data Quality transform to each Glue ETL job. Use Data Quality Definition Language (DQDL) to implement a ruleset that includes the data quality rules that need to be evaluated.
The Evaluate Data Quality transform is the native AWS Glue feature specifically designed for data quality checks. It uses Data Quality Definition Language (DQDL) to define rule sets with minimal implementation effort and integrates seamlessly into Glue Studio pipelines. Option A requires custom SQL, Option C and D require external libraries (PyDeequ, Great Expectations) that need custom transform development, adding implementation complexity.
Why the other options are wrong
- A. Writing custom SQL transforms for data quality is more cumbersome than using the dedicated data quality transform.
- C. PyDeequ requires custom transform development and additional library management, increasing implementation effort.
- D. Great Expectations is an external library requiring custom transform development and integration, not the least-effort approach.
Question 46
An investment company needs to manage and extract insights from a volume of semi-structured data that grows continuously. A data engineer needs to deduplicate the semi-structured data, remove records that are duplicates, and remove common misspellings of duplicates. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: A. Use the FindMatches feature of AWS Glue to remove duplicate records.
AWS Glue's FindMatches feature is purpose-built for deduplication and fuzzy matching of semi-structured data, specifically designed to handle misspellings and variations in duplicate records. It uses machine learning to identify and match similar records without requiring custom code, minimizing operational overhead. FindMatches is the AWS-native solution for this use case and requires minimal configuration. Athena window functions require custom SQL development, Neptune ML is overkill for deduplication, and DynamoDB global tables prevent duplicates during insertion but don't remediate existing duplicates.
Why the other options are wrong
- B. Window functions in Athena require custom SQL logic development and don't natively handle fuzzy matching for misspellings.
- C. Neptune ML and Gremlin are graph database tools, not designed for deduplication tasks and introduce unnecessary complexity.
- D. DynamoDB global tables prevent future duplicates but cannot remove existing duplicate records from semi-structured data.
Question 47
A company has a data warehouse that contains a table that is named Sales. The company stores the table in Amazon Redshift. The table includes a column that is named city_name. The company wants to query the table to find all rows that have a city_name that starts with "San" or "El". Which SQL query will meet this requirement?
Show answer and explanation
Correct answer: B. Select * from Sales where city_name ~ ‘^(San|El)*’;
The regex pattern '^(San|El)*' correctly matches city names that start with either 'San' or 'El'. The caret (^) anchors the match to the beginning of the string, and the pipe (|) operator provides the OR logic between the two options. The asterisk allows for any characters following the prefix.
Why the other options are wrong
- A. The dollar sign ($) anchors to the end of the string, not the beginning, so this matches cities ending with 'San' or 'El'.
- C. The ampersand (&) operator does not exist in regex; the pipe (|) is the correct OR operator, and the dollar sign is wrong.
- D. Uses the ampersand (&) which is not valid regex syntax, and the dollar sign ($) anchors to the end instead of the beginning.
Question 48
A data engineer must orchestrate a data pipeline that consists of one AWS Lambda function and one AWS Glue job. The solution must integrate with AWS services. Which solution will meet these requirements with the LEAST management overhead?
Show answer and explanation
Correct answer: A. Use an AWS Step Functions workflow that includes a state machine. Configure the state machine to run the Lambda function and then the AWS Glue job.
AWS Step Functions with a state machine provides serverless, managed orchestration specifically designed for AWS service integration. It requires minimal setup, just defining the state machine to run Lambda followed by Glue Job, and AWS handles all scaling and management. No infrastructure provisioning or third-party tools are needed.
Why the other options are wrong
- B. Apache Airflow on EC2 requires managing infrastructure, maintaining the Airflow service, and handling updates and patches.
- C. AWS Glue workflows are limited and not as flexible as Step Functions for complex orchestration scenarios.
- D. Apache Airflow on EKS requires Kubernetes infrastructure management, which adds significant operational overhead compared to Step Functions.
Question 49
A company has multiple applications that use datasets that are stored in an Amazon S3 bucket. The company has an ecommerce application that generates a dataset that contains personally identifiable information (PII). The company has an internal analytics application that does not require access to the PII. To comply with regulations, the company must not share PII unnecessarily. A data engineer needs to implement a solution that with redact PII dynamically, based on the needs of each application that accesses the dataset. Which solution will meet the requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: B. Create an S3 Object Lambda endpoint. Use the S3 Object Lambda endpoint to read data from the S3 bucket. Implement redaction logic within an S3 Object Lambda function to dynamically redact PII based on the needs of each application that accesses the data.
S3 Object Lambda enables dynamic transformation of data at read-time without creating multiple copies of datasets. By implementing redaction logic in an S3 Object Lambda function, the solution can apply different redaction levels based on the calling application's permissions or identity, providing compliance with least-privilege data access while maintaining a single source of truth with minimal operational overhead.
Why the other options are wrong
- A. Creating multiple dataset copies violates the principle of minimal data sharing and creates significant storage and maintenance overhead.
- C. Using AWS Glue to transform and create multiple copies requires managing multiple datasets and introduces unnecessary complexity and storage costs.
- D. API Gateway adds architectural complexity and requires maintaining separate API endpoints, making this solution more operationally intensive than Object Lambda.
Question 50
A media company uses software as a service (SaaS) applications to gather data by using third-party tools. The company needs to store the data in an Amazon S3 bucket. The company will use Amazon Redshift to perform analytics based on the data. Which AWS service or feature will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: B. Amazon AppFlow
Amazon AppFlow is a fully managed service that enables secure data integration from SaaS applications directly to AWS services like S3 and Redshift with minimal operational overhead. It provides pre-built connectors for popular SaaS tools, scheduled or event-driven data flows, and automatic data transformation capabilities, eliminating the need for custom integration code or infrastructure management.
Why the other options are wrong
- A. Amazon MSK is for streaming data from Kafka sources and adds operational complexity for SaaS integration.
- C. AWS Glue Data Catalog is a metadata repository; it does not ingest data from SaaS applications.
- D. Amazon Kinesis is for real-time streaming but requires custom integration code and more operational overhead than AppFlow for SaaS data sources.
Question 51
An online retail company stores Application Load Balancer (ALB) access logs in an Amazon S3 bucket. The company wants to use Amazon Athena to query the logs to analyze traffic patterns. A data engineer creates an unpartitioned table in Athena. As the amount of the data gradually increases, the response time for queries also increases. The data engineer wants to improve the query performance in Athena. Which solution will meet these requirements with the LEAST operational effort?
Show answer and explanation
Correct answer: B. Create an AWS Glue crawler that includes a classifier that determines the schema of all ALB access logs and writes the partition metadata to AWS Glue Data Catalog.
AWS Glue Crawler automatically discovers the schema of ALB access logs and creates partition metadata in the Data Catalog with minimal configuration. This solution provides the least operational effort as the crawler handles schema detection and partition generation automatically, allowing Athena to leverage partitions immediately without manual setup.
Why the other options are wrong
- A. AWS Glue job requires manual schema definition and programming effort to determine and write partition metadata
- C. Lambda transformation requires significant custom coding, data format conversion, and manual partitioning logic implementation
- D. Apache Hive and Lambda approach requires custom development and management of bucketed tables, adding operational complexity
Question 52
A company hosts its applications on Amazon EC2 instances. The company must use SSL/TLS connections that encrypt data in transit to communicate securely with AWS infrastructure that is managed by a customer. A data engineer needs to implement a solution to simplify the generation, distribution, and rotation of digital certificates. The solution must automatically renew and deploy SSL/TLS certificates. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: B. Use AWS Certificate Manager (ACM).
AWS Certificate Manager (ACM) automatically handles certificate generation, renewal, and deployment with no operational overhead. It integrates seamlessly with AWS services and eliminates manual certificate management entirely. Option A requires manual generation and rotation. Option C adds unnecessary complexity with custom scripts. Option D is for container service networking, not certificate management.
Why the other options are wrong
- A. Self-managed certificates require manual generation, distribution, and rotation, creating significant operational overhead.
- C. Custom automation scripts add complexity and maintenance burden compared to a fully managed service.
- D. Amazon ECS Service Connect is for service-to-service networking, not for SSL/TLS certificate lifecycle management.
Question 53
A data engineer needs Amazon Athena queries to finish faster. The data engineer notices that all the files the Athena queries use are currently stored in uncompressed .csv format. The data engineer also notices that users perform most queries by selecting a specific column. Which solution will MOST speed up the Athena query performance?
Show answer and explanation
Correct answer: C. Change the data format from .csv to Apache Parquet. Apply Snappy compression.
Apache Parquet is a columnar storage format optimized for analytical queries and selective column access. Since users primarily query specific columns, Parquet's columnar nature allows Athena to read only the required columns, significantly improving query speed. Combined with Snappy compression, it provides both fast query performance and reduced storage/bandwidth costs compared to uncompressed CSV.
Why the other options are wrong
- A. JSON is row-oriented and does not provide the columnar benefits needed for selective column queries.
- B. Snappy compression on CSV files still leaves the row-oriented format inefficient for column-specific queries.
- D. Gzip compression is effective for storage but does not address the inefficiency of querying specific columns from a row-oriented format.
Question 54
A marketing company uses Amazon S3 to store clickstream data. The company queries the data at the end of each day by using a SQL JOIN clause on S3 objects that are stored in separate buckets. The company creates key performance indicators (KPIs) based on the objects. The company needs a serverless solution that will give users the ability to query data by partitioning the data. The solution must maintain the atomicity, consistency, isolation, and durability (ACID) properties of the data. Which solution will meet these requirements MOST cost-effectively?
Show answer and explanation
Correct answer: C. Amazon Athena
Amazon Athena is the most cost-effective serverless solution for querying S3 data with SQL, supports partitioning natively to improve query performance and reduce costs, and maintains ACID properties through the AWS Glue Data Catalog integration. Athena charges only for data scanned, making it most cost-effective. Option A (S3 Select) is less powerful for complex queries and doesn't support partitioning well. Option B (Redshift Spectrum) requires a Redshift cluster, adding cost. Option D (EMR) requires cluster management and is not serverless.
Why the other options are wrong
- A. S3 Select is limited in query capability and doesn't optimize through partitioning
- B. Redshift Spectrum requires maintaining a Redshift cluster, increasing cost
- D. EMR requires cluster provisioning and management, not serverless, and is more expensive
Question 55
A company is migrating a legacy application to an Amazon S3 based data lake. A data engineer reviewed data that is associated with the legacy application. The data engineer found that the legacy data contained some duplicate information. The data engineer must identify and remove duplicate information from the legacy application data. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: B. Write an AWS Glue extract, transform, and load (ETL) job. Use the FindMatches machine learning (ML) transform to transform the data to perform data deduplication.
AWS Glue's FindMatches ML transform is purpose-built for deduplication and handles fuzzy matching of duplicate records automatically. It requires minimal configuration and no custom code, representing the least operational overhead. FindMatches learns patterns and handles complex duplicates that simple drop_duplicates() functions cannot detect.
Why the other options are wrong
- A. Custom Python with Pandas requires writing and maintaining custom code; drop_duplicates() only handles exact duplicates, not fuzzy matches.
- C. Custom Python with the dedupe library requires extensive tuning and training data preparation, adding operational complexity.
- D. While using Glue reduces some overhead compared to custom Python, the dedupe library still requires significant configuration and tuning compared to FindMatches.
Question 56
A company wants to use machine learning (ML) to perform analytics on data that is in an Amazon S3 data lake. The company has two data transformation requirements that will give consumers within the company the ability to create reports. The company must perform daily transformations on 300 GB of data that is in a variety format that must arrive in Amazon S3 at a scheduled time. The company must perform one-time transformations of terabytes of archived data that is in the S3 data lake. The company uses Amazon Managed Workflows for Apache Airflow (Amazon MWAA) Directed Acyclic Graphs (DAGs) to orchestrate processing. Which combination of tasks should the company schedule in the Amazon MWAA DAGs to meet these requirements MOST cost-effectively? (Choose two.)
Show answer and explanation
Correct answer: A, D. For daily incoming data, use AWS Glue crawlers to scan and identify the schema; For daily and archived data, use Amazon EMR to perform data transformations.
For daily incoming data, AWS Glue crawlers automatically scan and identify schema, which is cost-effective for structured transformation requirements. For archived terabytes of data, Amazon EMR provides a scalable, cost-effective solution for large-scale batch transformations. Glue crawlers integrate well with MWAA DAGs for orchestration, and EMR can be spun up on-demand for one-time transformations, minimizing costs.
Why the other options are wrong
- B. Athena is for querying, not for identifying schema for transformation pipelines.
- C. Redshift is expensive for daily 300GB transformations and better suited for analytics rather than ETL.
- E. SageMaker is for machine learning model training/inference, not general-purpose data transformations.
Question 57
A data engineer notices that Amazon Athena queries are held in a queue before the queries run. How can the data engineer prevent the queries from queueing?
Show answer and explanation
Correct answer: B. Configure provisioned capacity for an existing workgroup.
Provisioned capacity for Athena workgroups allocates dedicated compute resources that prevent query queueing by ensuring sufficient capacity is always available for query execution. This is AWS's recommended solution for preventing query queuing. Option A (query result limit) does not affect queue behavior. Option C (federated queries) expands data sources but doesn't prevent queuing. Option D (workgroup assignment) is configuration for access control, not queue management.
Why the other options are wrong
- A. Query result limits affect output size, not queue behavior
- C. Federated queries access external data sources but don't prevent queue delays
- D. Workgroup assignment controls permissions but does not prevent query queueing
Question 58
A retail company uses AWS Glue for extract, transform, and load (ETL) operations on a dataset that contains information about customer orders. The company wants to implement specific validation rules to ensure data accuracy and consistency. Which solution will meet these requirements?
Show answer and explanation
Correct answer: B. Create custom AWS Glue Data Quality rulesets to define specific data quality checks.
AWS Glue Data Quality rulesets allow companies to define custom, specific validation rules tailored to their data accuracy and consistency requirements. This provides explicit control over what constitutes valid data and enables monitoring against defined business rules, which is exactly what the question asks for.
Why the other options are wrong
- A. Job bookmarks track processed data for incremental runs, not for implementing validation rules.
- C. Built-in transforms are standard validations, not custom specific rules tailored to business requirements.
- D. Data Catalog maintains metadata and schema but does not perform data quality validation checks.
Question 59
A retail company uses an Amazon Redshift data warehouse and an Amazon S3 bucket. The company ingests retail order data into the S3 bucket every day. The company stores all order data at a single path within the S3 bucket. The data has more than 100 columns. The company ingests the order data from a third-party application that generates more than 30 files in CSV format every day. Each CSV file is between 50 and 70 MB in size. The company uses Amazon Redshift Spectrum to run queries that select sets of columns. Users aggregate metrics based on daily orders. Recently, users have reported that the performance of the queries has degraded. A data engineer must resolve the performance issues for the queries. Which combination of steps will meet this requirement with LEAST developmental effort? (Choose two.)
Show answer and explanation
Correct answer: A, C. Configure the third-party application to create the files in a columnar format; Partition the order data in the S3 bucket based on order date.
Converting CSV files to a columnar format (A) dramatically improves Spectrum query performance by reducing data scanned when selecting specific columns from 100+ column datasets. Partitioning by date (C) enables partition pruning, allowing Spectrum to skip entire partitions not matching the query filter, significantly reducing data scanned for daily aggregations. Together, these address the performance degradation with minimal development effort. Converting to JSON (D) doesn't provide columnar benefits. Consolidating files (B) reduces query overhead but is secondary to columnar format and partitioning. SUPER type (E) is unnecessary complexity.
Why the other options are wrong
- B. Consolidating multiple files into one daily file provides minor improvement but doesn't address the core columnar format and partitioning needs.
- D. JSON format is also row-oriented like CSV and doesn't provide the columnar performance benefits needed for selective column queries.
- E. Using SUPER type columns adds complexity without addressing the fundamental performance issues of row-oriented CSV format and missing partitions.
Question 60
A company has implemented a lake house architecture in Amazon Redshift. The company needs to give users the ability to authenticate into Redshift query editor by using a third-party identity provider (IdP). A data engineer must set up the authentication mechanism. What is the first step the data engineer should take to meet this requirement?
Show answer and explanation
Correct answer: B. Register the third-party IdP as an identity provider from within Amazon Redshift.
To enable third-party IdP authentication for Redshift query editor, the first step is to register the IdP as an identity provider from within Amazon Redshift itself using the Redshift console or API. This establishes the trust relationship between Redshift and the external IdP, allowing users to authenticate using their IdP credentials. Once registered in Redshift, users can then use federated authentication to access the query editor.
Why the other options are wrong
- A. While this might seem plausible, IdP registration is not done in cluster configuration settings but rather through Redshift's identity provider management interface.
- C. AWS Secrets Manager is used for storing credentials, not for registering IdPs; this approach does not enable federated authentication.
- D. AWS Certificate Manager manages certificates, not IdP registration; it is not the appropriate service for this use case.
Question 61
A telecommunications company collects network usage data throughout each day at a rate of several thousand data points each second. The company runs an application to process the usage data in real time. The company aggregates and stores the data in an Amazon Aurora DB instance. Sudden drops in network usage usually indicate a network outage. The company must be able to identify sudden drops in network usage so the company can take immediate remedial actions. Which solution will meet this requirement with the LEAST latency?
Show answer and explanation
Correct answer: B. Modify the processing application to publish the data to an Amazon Kinesis data stream. Create an Amazon Managed Service for Apache Flink (previously known as Amazon Kinesis Data Analytics) application to detect drops in network usage.
Amazon Managed Service for Apache Flink provides the LEAST latency for real-time anomaly detection because it processes data streams continuously in-memory as it arrives. By consuming from a Kinesis data stream, Flink can detect sudden drops in network usage with minimal delay, enabling immediate remedial action. This is purpose-built for real-time event processing, unlike batch-oriented query approaches.
Why the other options are wrong
- A. Lambda invoked every minute introduces up to 60 seconds of latency, which is unacceptable for detecting sudden drops in real-time
- C. Querying DynamoDB every minute also introduces significant latency compared to continuous stream processing
- D. Database Activity Streams is for database-level monitoring and auditing, not for processing application data in real-time
Question 62
An insurance company stores transaction data that the company compressed with gzip. The company needs to query the transaction data for occasional audits. Which solution will meet this requirement in the MOST cost-effective way?
Show answer and explanation
Correct answer: B. Store the data in Amazon S3. Use Amazon S3 Select to query the data.
Amazon S3 Select can query gzip-compressed data directly without decompression, providing a cost-effective solution for occasional audits. S3 is the most economical storage option, and S3 Select charges only for the data scanned, making it ideal for infrequent queries on compressed transaction data. This avoids the higher storage and retrieval costs of Glacier tiers.
Why the other options are wrong
- A. Glacier Flexible Retrieval has retrieval time delays and higher costs for occasional queries; Glacier Select also incurs retrieval fees.
- C. Athena requires data to be in standard format and incurs scanning charges, making it less cost-effective than S3 Select for occasional audits.
- D. Glacier Instant Retrieval still costs more than S3 for storage and retrieval, and adds unnecessary complexity.
Question 63
A company stores datasets in JSON format and .csv format in an Amazon S3 bucket. The company has Amazon RDS for Microsoft SQL Server databases, Amazon DynamoDB tables that are in provisioned capacity mode, and an Amazon Redshift cluster. A data engineering team must develop a solution that will give data scientists the ability to query all data sources by using syntax similar to SQL. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: A. Use AWS Glue to crawl the data sources. Store metadata in the AWS Glue Data Catalog. Use Amazon Athena to query the data. Use SQL for structured data sources. Use PartiQL for data that is stored in JSON format.
AWS Glue with the Data Catalog provides schema detection for diverse data formats and sources with minimal operational overhead. Amazon Athena can query all these sources using standard SQL syntax, while PartiQL handles JSON-specific queries. This approach requires no data transformation or ETL jobs, reducing operational complexity while meeting the requirement for SQL-like query access across all data sources.
Why the other options are wrong
- B. Redshift Spectrum requires data to be in specific formats and adds operational overhead compared to the simpler Athena approach.
- C. This option requires additional transformation steps and creates duplicate data, increasing operational overhead unnecessarily.
- D. Lake Formation is more complex and operationally demanding than simple Glue crawling with Athena, and requires unnecessary data transformation to Parquet.
Question 64
A data engineer needs to securely transfer 5 TB of data from an on-premises data center to an Amazon S3 bucket. Approximately 5% of the data changes every day. Updates to the data need to be regularly proliferated to the S3 bucket. The data includes files that are in multiple formats. The data engineer needs to automate the transfer process and must schedule the process to run periodically. Which AWS service should the data engineer use to transfer the data in the MOST operationally efficient way?
Show answer and explanation
Correct answer: A. AWS DataSync
AWS DataSync is purpose-built for automated, scheduled data transfer between on-premises systems and AWS. It efficiently handles incremental changes (5% daily updates), supports multiple file formats natively, and provides automated scheduling capabilities. DataSync includes built-in data validation, encryption, and bandwidth optimization, making it the most operationally efficient solution for this use case.
Why the other options are wrong
- B. AWS Glue is designed for ETL data transformation, not efficient bulk data transfer and synchronization.
- C. Direct Connect is a network connectivity service, not a data transfer service, and would still require additional tools for automation.
- D. S3 Transfer Acceleration optimizes upload speed but does not provide scheduling, automation, or incremental sync capabilities.
Question 65
A company needs to build a data lake in AWS. The company must provide row-level data access and column-level data access to specific teams. The teams will access the data by using Amazon Athena, Amazon Redshift Spectrum, and Apache Hive from Amazon EMR. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: D. Use Amazon S3 for data lake storage. Use AWS Lake Formation to restrict data access by rows and columns. Provide data access through AWS Lake Formation.
AWS Lake Formation provides fine-grained access control with both row-level and column-level permissions natively integrated with S3, Athena, Redshift Spectrum, and EMR. It is specifically designed for data lake scenarios and manages permissions centrally with minimal operational overhead compared to managing policies manually or through third-party tools.
Why the other options are wrong
- A. S3 access policies cannot enforce row-level or column-level restrictions; they operate at object level only.
- B. Apache Ranger requires EMR cluster management and only works with EMR/Hive, not Athena or Redshift Spectrum directly.
- C. Redshift is a data warehouse, not optimal for data lake storage; Redshift security policies don't provide the flexibility needed for multi-tool access patterns.
Question 66
A company saves customer data to an Amazon S3 bucket. The company uses server-side encryption with AWS KMS keys (SSE-KMS) to encrypt the bucket. The dataset includes personally identifiable information (PII) such as social security numbers and account details. Data that is tagged as PII must be masked before the company uses customer data for analysis. Some users must have secure access to the PII data during the pre-processing phase. The company needs a low-maintenance solution to mask and secure the PII data throughout the entire engineering pipeline. Which combination of solutions will meet these requirements? (Choose two.)
Show answer and explanation
Correct answer: A, D. Use AWS Glue DataBrew to perform extract, transform, and load (ETL) tasks that mask the PII data before analysis; Use AWS Identity and Access Management (IAM) to manage permissions and to control access to the PII data.
Option A (AWS Glue DataBrew) provides built-in PII masking capabilities and is specifically designed for low-maintenance ETL tasks that can mask sensitive data before analysis. Option D (IAM) provides the access control and permission management needed to secure the PII data and ensure only authorized users can access it during the pre-processing phase. Together, these solutions provide both the masking functionality and the security controls required. Option B monitors but doesn't secure or mask. Option C discovers PII but doesn't mask it. Option E requires custom development contrary to the low-maintenance requirement.
Why the other options are wrong
- B. GuardDuty monitors access patterns but does not mask PII data or provide data protection.
- C. Macie discovers and classifies PII but does not mask the data or control access to it.
- E. Custom scripts require high maintenance and development effort, contradicting the low-maintenance requirement.
Question 67
A company stores customer data that contains personally identifiable information (PII) in an Amazon Redshift cluster. The company's marketing, claims, and analytics teams need to be able to access the customer data. The marketing team should have access to obfuscated claim information but should have full access to customer contact information. The claims team should have access to customer information for each claim that the team processes. The analytics team should have access only to obfuscated PII data. Which solution will enforce these data access requirements with the LEAST administrative overhead?
Show answer and explanation
Correct answer: C. Create a separate Amazon Redshift database role for each team. Define masking policies that apply for each team separately. Attach appropriate masking policies to each team role.
Redshift dynamic masking policies allow role-based access with automatic data obfuscation applied at query time. Creating a database role per team with attached masking policies enforces granular access control where each team sees only permitted data with appropriate obfuscation, minimizing administrative overhead by centralizing policy management within a single cluster.
Why the other options are wrong
- A. Maintaining separate clusters for each team requires significant operational overhead and data duplication.
- B. Views require manual creation and maintenance of multiple view definitions, and do not provide dynamic masking for sensitive fields.
- D. Moving to Lake Formation on S3 introduces migration complexity and requires rebuilding data infrastructure, increasing overhead.
Question 68
A company analyzes data in a data lake every quarter to perform inventory assessments. A data engineer uses AWS Glue DataBrew to detect any personally identifiable formation (PII) about customers within the data. The company's privacy policy considers some custom categories of information to be PII. However, the categories are not included in standard DataBrew data quality rules. The data engineer needs to modify the current process to scan for the custom PII categories across multiple datasets within the data lake. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: B. Implement custom data quality rules in DataBrew. Apply the custom rules across datasets.
AWS Glue DataBrew supports custom data quality rules that can be defined to detect company-specific PII categories. These custom rules can be applied across multiple datasets within the data lake with minimal operational overhead. This is a native DataBrew feature designed exactly for this use case.
Why the other options are wrong
- A. Manual review is not scalable and has high operational overhead.
- C. While possible, developing and maintaining custom Python scripts adds unnecessary complexity and operational overhead compared to built-in custom rules.
- D. Implementing regex patterns in ETL operations is manual, requires custom development, and is harder to maintain than DataBrew's custom rule framework.
Question 69
A data engineer is configuring Amazon SageMaker Studio to use AWS Glue interactive sessions to prepare data for machine learning (ML) models. The data engineer receives an access denied error when the data engineer tries to prepare the data by using SageMaker Studio. Which change should the engineer make to gain access to SageMaker Studio?
Show answer and explanation
Correct answer: B. Add a policy to the data engineer's IAM user that includes the sts:AssumeRole action for the AWS Glue and SageMaker service principals in the trust policy.
For SageMaker Studio to use AWS Glue interactive sessions, the IAM user needs permission to assume a role that has access to both AWS Glue and SageMaker resources. This requires adding a policy that includes the sts:AssumeRole action with trust relationships for both AWS Glue and SageMaker service principals, allowing the session to operate with the appropriate permissions.
Why the other options are wrong
- A. AWSGlueServiceRole is a service role for Glue itself, not a permission an IAM user needs to assume to use Glue sessions.
- C. AmazonSageMakerFullAccess alone does not grant permission to assume roles or interact with Glue services.
- D. sts:AddAssociation is not the correct action; sts:AssumeRole is what's needed to use cross-service integration.
Question 70
A technology company currently uses Amazon Kinesis Data Streams to collect log data in real time. The company wants to use Amazon Redshift for downstream real-time queries and to enrich the log data. Which solution will ingest data into Amazon Redshift with the LEAST operational overhead?
Show answer and explanation
Correct answer: D. Use Amazon Redshift streaming ingestion from Kinesis Data Streams and to present data as a materialized view.
Amazon Redshift streaming ingestion from Kinesis Data Streams allows data to flow directly from Kinesis into Redshift with minimal latency and operational overhead. The streaming data is presented as a materialized view, enabling real-time queries and enrichment without intermediate storage layers. This is the most operationally efficient approach as it eliminates ETL processes and external services, maintaining a direct data pipeline.
Why the other options are wrong
- A. Firehose delivery requires buffering and batch processing, adding latency compared to direct streaming ingestion.
- B. Adding S3 as an intermediate layer and scheduled loading introduces significant latency and operational complexity.
- C. Managed Flink adds another service layer for transformation; direct streaming ingestion is simpler and more efficient.
Question 71
A data engineer has a one-time task to read data from objects that are in Apache Parquet format in an Amazon S3 bucket. The data engineer needs to query only one column of the data. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: B. Use S3 Select to write a SQL SELECT statement to retrieve the required column from the S3 objects.
S3 Select is purpose-built for querying specific columns from objects in S3 without loading the entire dataset into memory. For a one-time task requiring minimal operational overhead, S3 Select with a SQL SELECT statement is the most efficient solution, it requires no infrastructure setup, no function deployment, and no crawler configuration. It directly queries the Parquet file and returns only the requested column, making it the least overhead option.
Why the other options are wrong
- A. Lambda with pandas requires writing and managing function code, installing dependencies, and handling data loading inefficiently when only one column is needed.
- C. DataBrew is designed for data preparation workflows and requires project setup, making it overkill for a simple one-time column extraction.
- D. Crawlers and Athena require metadata cataloging setup, Glue job execution, and ongoing resource management, significantly more overhead than S3 Select.
Question 72
A company is planning to use a provisioned Amazon EMR cluster that runs Apache Spark jobs to perform big data analysis. The company requires high reliability. A big data team must follow best practices for running cost-optimized and long-running workloads on Amazon EMR. The team must find a solution that will maintain the company's current level of performance. Which combination of resources will meet these requirements MOST cost-effectively? (Choose two.)
Show answer and explanation
Correct answer: B, D. Use Amazon S3 as a persistent data store; Use Graviton instances for core nodes and task nodes.
Amazon S3 is the ideal persistent data store for EMR because it's cost-effective, scalable, and separates compute from storage, allowing cluster flexibility. Graviton instances are ARM-based processors that offer better price-to-performance ratios compared to x86, reducing costs while maintaining performance for Spark workloads. Together, these choices optimize costs for long-running workloads while maintaining reliability.
Why the other options are wrong
- A. HDFS is tied to the cluster and disappears when the cluster terminates, unsuitable for persistent long-term storage.
- C. x86-based instances are more expensive than Graviton instances with comparable performance.
- E. Using Spot Instances for primary nodes compromises reliability; Spot instances can be interrupted, causing cluster failures.
Question 73
A company wants to migrate data from an Amazon RDS for PostgreSQL DB instance in the eu-east-1 Region of an AWS account named Account_A. The company will migrate the data to an Amazon Redshift cluster in the eu-west-1 Region of an AWS account named Account_B. Which solution will give AWS Database Migration Service (AWS DMS) the ability to replicate data between two data stores?
Show answer and explanation
Correct answer: A. Set up an AWS DMS replication instance in Account_B in eu-west-1.
The DMS replication instance should be placed in the destination region (eu-west-1) and account (Account_B) where the target Redshift cluster resides. This minimizes network latency and costs for writing data to the destination. The replication instance connects to the source database across the Direct Connect link and writes to the destination locally. Option B places it in the source region. Option C creates unnecessary infrastructure in a third account. Option D places it in the source account rather than destination.
Why the other options are wrong
- B. Placing the replication instance in the source region (eu-east-1) adds latency for data transfer to the destination
- C. Creating a replication instance in a new account adds unnecessary complexity and governance overhead
- D. Placing it in the source account (Account_A) requires cross-account networking and adds latency to destination writes
Question 74
A marketing company collects clickstream data. The company sends the clickstream data to Amazon Kinesis Data Firehose and stores the clickstream data in Amazon S3. The company wants to build a series of dashboards that hundreds of users from multiple departments will use. The company will use Amazon QuickSight to develop the dashboards. The company wants a solution that can scale and provide daily updates about clickstream activity. Which combination of steps will meet these requirements MOST cost-effectively? (Choose two.)
Show answer and explanation
Correct answer: B, E. Use Amazon Athena to query the clickstream data; Access the query data through QuickSight SPICE (Super-fast, Parallel, In-memory Calculation Engine). Configure a daily refresh for the dataset.
Amazon Athena can query clickstream data directly from S3 in a cost-effective manner, charging only for data scanned. QuickSight SPICE with daily refresh provides fast in-memory analysis for hundreds of concurrent users while keeping costs down by caching data rather than querying source systems repeatedly. Together, these provide scalability and cost-effectiveness for multi-departmental dashboards.
Why the other options are wrong
- A. Redshift requires cluster provisioning and management, making it more expensive than Athena for this use case.
- C. S3 analytics is not a query service and cannot be used to query clickstream data for dashboards.
- D. Direct SQL queries bypass caching and require continuous querying, increasing costs and reducing performance for hundreds of users.
Question 75
A gaming company uses Amazon Kinesis Data Streams to collect clickstream data. The company uses Amazon Data Firehose delivery streams to store the data in JSON format in Amazon S3. Data scientists at the company use Amazon Athena to query the most recent data to obtain business insights. The company wants to reduce Athena costs but does not want to recreate the data pipeline. Which solution will meet these requirements with the LEAST management effort?
Show answer and explanation
Correct answer: A. Change the Firehose output format to Apache Parquet. Provide a custom S3 object YYYYMMDD prefix expression and specify a large buffer size. For the existing data, create an AWS Glue extract, transform, and load (ETL) job. Configure the ETL job to combine small JSON files, convert the JSON files to large Parquet files, and add the YYYYMMDD prefix. Use the ALTER TABLE ADD PARTITION statement to reflect the partition on the existing Athena table.
Converting Firehose output to Parquet format directly reduces Athena query costs since Parquet is columnar and compressed. Configuring Firehose with custom S3 prefixes and large buffer sizes optimizes file organization for querying. For existing JSON data, a Glue ETL job can perform a one-time conversion. This approach requires minimal ongoing management and uses native AWS services without introducing new components like EMR or Flink.
Why the other options are wrong
- B. Launching an EMR cluster daily adds significant operational overhead and cost compared to serverless alternatives.
- C. Adding Managed Flink as an intermediary introduces unnecessary complexity and operational overhead for what can be achieved with Firehose and Glue.
- D. Running both Lambda transformation and parallel Glue jobs introduces redundancy and complexity; Firehose can handle the conversion directly.
Question 76
A company has developed several AWS Glue extract, transform, and load (ETL) jobs to validate and transform data from Amazon S3. The ETL jobs load the data into Amazon RDS for MySQL in batches once every day. The ETL jobs use a DynamicFrame to read the S3 data. The ETL jobs currently process all the data that is in the S3 bucket. However, the company wants the jobs to process only the daily incremental data. Which solution will meet this requirement with the LEAST coding effort?
Show answer and explanation
Correct answer: B. Enable job bookmarks for the ETL jobs to update the state after a run to keep track of previously processed data.
Job bookmarks are a built-in Glue feature that automatically tracks the state of processed data and ensures only incremental data is processed on subsequent runs. This requires minimal code changes, just enabling the feature, making it the least-effort solution for processing daily incremental data instead of all data.
Why the other options are wrong
- A. Implementing custom file status tracking in DynamoDB requires significant coding effort and manual state management.
- C. Job metrics in CloudWatch track performance data but do not provide state tracking for determining which data has been processed.
- D. Deleting processed objects is destructive and inefficient, introducing operational risk and wasted effort compared to using job bookmarks.
Question 77
A mobile gaming company wants to capture data from its gaming app. The company wants to make the data available to three internal consumers of the data. The data records are approximately 20 KB in size. The company wants to achieve optimal throughput from each device that runs the gaming app. Additionally, the company wants to develop an application to process data streams. The stream-processing application must have dedicated throughput for each internal consumer. Which solution will meet these requirements?
Show answer and explanation
Correct answer: A. Configure the mobile app to call the PutRecords API operation to send data to Amazon Kinesis Data Streams. Use the enhanced fan-out feature with a stream for each internal consumer.
Sending data to Kinesis Data Streams via PutRecords (A) with enhanced fan-out provides dedicated throughput for each internal consumer. Enhanced fan-out ensures that each consumer receives its own dedicated connection and throughput allocation, eliminating contention. Multiple streams (one per consumer) with enhanced fan-out fully addresses the requirement for dedicated throughput per consumer while optimizing device-side throughput. Firehose (B) is for data delivery to storage endpoints, not stream processing. KPL with Firehose (C) combines incompatible components. EC2 with auto scaling (D) creates operational overhead without the throughput guarantees of enhanced fan-out.
Why the other options are wrong
- B. Firehose is designed for delivery to storage targets, not for stream processing with multiple consumer requirements.
- C. KPL is designed for Kinesis Streams, not Firehose; using it with Firehose is architecturally incorrect.
- D. Managing EC2 instances adds operational overhead and doesn't provide the throughput guarantees of enhanced fan-out.
Question 78
A data engineer must orchestrate a series of Amazon Athena queries that will run every day. Each query can run for more than 15 minutes. Which combination of steps will meet these requirements MOST cost-effectively? (Choose two.)
Show answer and explanation
Correct answer: A, B. Use an AWS Lambda function and the Athena Boto3 client start_query_execution API call to invoke the Athena queries programmatically; Create an AWS Step Functions workflow and add two states. Add the first state before the Lambda function. Configure the second state as a Wait state to periodically check whether the Athena query has finished using the Athena Boto3 get_query_execution API call. Configure the workflow to invoke the next query when the current query has finished running.
Option A provides cost-effective programmatic invocation of Athena queries using Lambda and the Boto3 API. Option B adds Step Functions with a Wait state to handle polling and orchestration without consuming Lambda run time during query execution, which is crucial for 15+ minute queries. This combination avoids wasting compute resources on polling loops and provides serverless, event-driven orchestration at minimal cost. Together, they create an efficient, cost-effective solution that doesn't waste time or money on sleep timers or unnecessary Glue overhead.
Why the other options are wrong
- C. Glue Python shell jobs incur compute costs even during query waits, making them less cost-effective than Lambda + Step Functions.
- D. Sleep timers in Python scripts waste compute resources and are inefficient for long-running queries; Step Functions Wait states are purpose-built for this.
- E. MWAA adds unnecessary complexity and overhead; Step Functions is simpler and more cost-effective for this use case.
Question 79
A company receives a daily file that contains customer data in .xls format. The company stores the file in Amazon S3. The daily file is approximately 2 GB in size. A data engineer concatenates the column in the file that contains customer first names and the column that contains customer last names. The data engineer needs to determine the number of distinct customers in the file. Which solution will meet this requirement with the LEAST operational effort?
Show answer and explanation
Correct answer: D. Use AWS Glue DataBrew to create a recipe that uses the COUNT_DISTINCT aggregate function to calculate the number of distinct customers.
AWS Glue DataBrew is a visual data preparation tool specifically designed for this use case. It can load the .xls file, concatenate the name columns through a recipe, and use the COUNT_DISTINCT aggregate function to calculate distinct customers without writing code. This requires minimal operational effort and no cluster management. Option A requires writing Spark code and managing a notebook; option B requires crawler setup and writing SQL queries; option C requires EMR cluster provisioning and Spark coding.
Why the other options are wrong
- A. Requires writing and debugging Apache Spark code in a notebook, more effort than using DataBrew's visual interface.
- B. Requires crawler setup, Data Catalog configuration, and writing SQL queries; more steps than DataBrew.
- C. Requires provisioning EMR Serverless and writing Spark code, more complex than DataBrew's visual approach.
Question 80
A lab uses IoT sensors to monitor humidity, temperature, and pressure for a project. The sensors send 100 KB of data every 10 seconds. A downstream process will read the data from an Amazon S3 bucket every 30 seconds. Which solution will deliver the data to the S3 bucket with the LEAST latency?
Show answer and explanation
Correct answer: C. Use Amazon Kinesis Data Streams and call the Kinesis Client Library to deliver the data to the S3 bucket. Use a 5 second buffer interval from an application.
Using Kinesis Data Streams with the Kinesis Client Library and a 5-second buffer interval minimizes latency by allowing the application to push data to S3 as soon as the buffer fills, matching the 10-second sensor interval and 30-second downstream processing window. This avoids Firehose's default 300-second buffer and unnecessary stream overhead while keeping data fresh.
Why the other options are wrong
- A. Kinesis Data Firehose's default buffer interval is 300 seconds, introducing excessive latency for near real-time requirements.
- B. Using only Kinesis Data Streams requires manual S3 delivery logic and does not address buffering strategy for latency optimization.
- D. Managed Apache Flink adds unnecessary complexity for this simple data delivery use case and does not provide latency advantages over the simpler solution.
Question 81
A security company stores IoT data that is in JSON format in an Amazon S3 bucket. The data structure can change when the company upgrades the IoT devices. The company wants to create a data catalog that includes the IoT data. The company's analytics department will use the data catalog to index the data. Which solution will meet these requirements MOST cost-effectively?
Show answer and explanation
Correct answer: A. Create an AWS Glue Data Catalog. Configure an AWS Glue Schema Registry. Create a new AWS Glue workload to orchestrate the ingestion of the data that the analytics department will use into Amazon Redshift Serverless.
AWS Glue Data Catalog combined with AWS Glue Schema Registry is the most cost-effective serverless solution for managing evolving JSON data structures. The Schema Registry handles schema evolution automatically, the Data Catalog provides indexing, and the orchestration into Redshift Serverless (also serverless) keeps costs minimal by paying only for what's used. No infrastructure provisioning is required.
Why the other options are wrong
- B. Amazon Redshift provisioned cluster requires ongoing per-node costs regardless of utilization, and Spectrum adds complexity without the flexibility of a pure data catalog approach.
- C. While Athena is serverless and cost-effective for querying, it doesn't create a managed data catalog with schema management capabilities for evolving structures.
- D. This combines Glue Data Catalog with Lambda UDFs and Step Functions, adding unnecessary complexity and costs compared to the simpler integrated solution in option A.
Question 82
A data engineer must ingest a source of structured data that is in .csv format into an Amazon S3 data lake. The .csv files contain 15 columns. Data analysts need to run Amazon Athena queries on one or two columns of the dataset. The data analysts rarely query the entire file. Which solution will meet these requirements MOST cost-effectively?
Show answer and explanation
Correct answer: D. Create an AWS Glue extract, transform, and load (ETL) job to read from the .csv structured data source. Configure the job to write the data into the data lake in Apache Parquet format.
Apache Parquet is a columnar storage format that is ideal for analytical queries on a subset of columns. Since data analysts rarely query the entire file and only access one or two columns out of 15, Parquet's columnar structure allows Athena to read only the required columns, significantly reducing I/O costs and query latency compared to row-based formats like CSV or JSON. Avro is also columnar but less optimized for Athena queries than Parquet.
Why the other options are wrong
- A. CSV format requires reading all 15 columns even when querying just one or two, wasting I/O and increasing costs.
- B. JSON format is row-based and also requires reading entire records, making it inefficient for selective column access.
- C. While Avro is a good format, Parquet is more optimized for analytical queries with selective column access in Athena.
Question 83
A data engineer runs Amazon Athena queries on data that is in an Amazon S3 bucket. The Athena queries use AWS Glue Data Catalog as a metadata table. The data engineer notices that the Athena query plans are experiencing a performance bottleneck. The data engineer determines that the cause of the performance bottleneck is the large number of partitions that are in the S3 bucket. The data engineer must resolve the performance bottleneck and reduce Athena query planning time. Which solutions will meet these requirements? (Choose two.)
Show answer and explanation
Correct answer: A, C. Create an AWS Glue partition index. Enable partition filtering; Use Athena partition projection based on the S3 bucket prefix.
Partition indexes with filtering (Option A) reduce query planning time by quickly identifying relevant partitions without scanning metadata for all partitions. Partition projection (Option C) uses S3 prefix patterns to avoid Glue Catalog lookups entirely, providing the fastest partition discovery. Both address the root cause of slow query planning from excessive partitions.
Why the other options are wrong
- B. Bucketing data improves query execution but does not reduce partition metadata discovery time during planning.
- D. Converting to Parquet improves query execution speed but does not address the partition planning bottleneck.
- E. S3DistCP combines objects but does not reduce the number of partitions or improve partition discovery speed.
Question 84
A retail company stores data from a product lifecycle management (PLM) application in an on-premises MySQL database. The PLM application frequently updates the database when transactions occur. The company wants to gather insights from the PLM application in near real time. The company wants to integrate the insights with other business datasets and to analyze the combined dataset by using an Amazon Redshift data warehouse. The company has already established an AWS Direct Connect connection between the on-premises infrastructure and AWS. Which solution will meet these requirements with the LEAST development effort?
Show answer and explanation
Correct answer: B. Run a full load plus CDC task in AWS Database Migration Service (AWS DMS) to continuously replicate the MySQL database changes. Set Amazon Redshift as the destination for the task.
AWS DMS with Change Data Capture (CDC) is purpose-built for continuous replication of database changes to Redshift in near real-time, requiring minimal development effort. DMS handles the complexity of capturing and replicating MySQL changes automatically. Option A requires scheduling and full table scans, not near real-time updates. Option C requires building custom connectors with AppFlow SDK, which increases development effort. Option D (DataSync) is for file synchronization, not database change capture.
Why the other options are wrong
- A. Scheduled Glue jobs provide periodic updates only, not near real-time continuous replication
- C. Building custom AppFlow connectors requires significant development effort beyond requirements
- D. DataSync is designed for file-based synchronization, not transactional database change capture
Question 85
A manufacturing company wants to collect data from sensors. A data engineer needs to implement a solution that ingests sensor data in near real time. The solution must store the data to a persistent data store. The solution must store the data in nested JSON format. The company must have the ability to query from the data store with a latency of less than 10 milliseconds. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: C. Use Amazon Kinesis Data Streams to capture the sensor data. Store the data in Amazon DynamoDB for querying.
Amazon Kinesis Data Streams provides near-real-time ingestion of sensor data with minimal operational overhead, and Amazon DynamoDB offers sub-10-millisecond latency queries on nested JSON data stored natively. DynamoDB is designed for this use case with automatic scaling and no server management. Together they provide a fully managed, serverless solution meeting all requirements.
Why the other options are wrong
- A. Self-hosted Kafka requires cluster management and operational overhead; S3 query latency is minutes to seconds, not under 10ms.
- B. Lambda is event-driven and adds latency; S3 query latency does not meet the sub-10ms requirement.
- D. SQS is for queuing, not real-time streaming; RDS with AWS Glue adds complexity and does not provide sub-10ms latency at scale.
Question 86
A company receives a data file from a partner each day in an Amazon S3 bucket. The company uses a daily AWS Glue extract, transform, and load (ETL) pipeline to clean and transform each data file. The output of the ETL pipeline is written to a CSV file named Daily.csv in a second S3 bucket. Occasionally, the daily data file is empty or is missing values for required fields. When the file is missing data, the company can use the previous day’s CSV file. A data engineer needs to ensure that the previous day's data file is overwritten only if the new daily file is complete and valid. Which solution will meet these requirements with the LEAST effort?
Show answer and explanation
Correct answer: B. Configure the AWS Glue ETL pipeline to use AWS Glue Data Quality rules. Develop rules in Data Quality Definition Language (DQDL) to check for missing values in required fields and empty files.
AWS Glue Data Quality rules defined in DQDL can automatically validate incoming data for completeness and detect empty files or missing required fields. The ETL pipeline can be configured to only write output if validation passes, ensuring the previous day's file is overwritten only when new data is valid. This provides automated, built-in validation with minimal effort.
Why the other options are wrong
- A. Lambda would require custom code development and doesn't integrate as naturally with Glue pipelines.
- C. Filling missing values with most common values is data transformation, not validation; it doesn't prevent incomplete data from overwriting good data.
- D. Running Athena queries requires manual setup and doesn't integrate with the automated pipeline; it's a reactive rather than preventive approach.
Question 87
A company is building a data stream processing application. The application runs in an Amazon Elastic Kubernetes Service (Amazon EKS) cluster. The application stores processed data in an Amazon DynamoDB table. The company needs the application containers in the EKS cluster to have secure access to the DynamoDB table. The company does not want to embed AWS credentials in the containers. Which solution will meet these requirements?
Show answer and explanation
Correct answer: B. Attach an IAM role to the EKS worker nodes, Grant the IAM role access to DynamoDUse the IAM role to set up IAM roles service accounts (IRSA) functionality.
IAM Roles for Service Accounts (IRSA) is the AWS-native solution that allows EKS containers to assume an IAM role without embedding credentials. By attaching an IAM role to EKS worker nodes and configuring IRSA, containers can securely access DynamoDB through temporary credentials managed by the AWS Security Token Service, eliminating the need to store or embed credentials.
Why the other options are wrong
- A. Storing credentials in S3 and retrieving them embeds credential access in containers, which is less secure than using IRSA.
- C. Using environment variables to store IAM access keys embeds long-lived credentials in containers, violating security best practices.
- D. Storing credentials in Kubernetes secrets mounted as volumes still embeds static credentials in the cluster, which is less secure than temporary role-based credentials.
Question 88
A healthcare company uses Amazon Kinesis Data Streams to stream real-time health data from wearable devices, hospital equipment, and patient records. A data engineer needs to find a solution to process the streaming data. The data engineer needs to store the data in an Amazon Redshift Serverless warehouse. The solution must support near real-time analytics of the streaming data and the previous day's data. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: B. Use the streaming ingestion feature of Amazon Redshift.
Amazon Redshift's streaming ingestion feature allows direct ingestion of data from Kinesis Data Streams with minimal operational overhead. It provides near real-time analytics without requiring intermediate services like Firehose or S3, and works seamlessly with Redshift Serverless. This is the most direct and operationally efficient solution for this use case.
Why the other options are wrong
- A. Kinesis Data Firehose adds an extra layer and operational complexity compared to direct Redshift streaming ingestion.
- C. Using S3 with COPY commands introduces additional latency and operational overhead in the ETL pipeline.
- D. Aurora zero-ETL integration is designed for relational databases, not for Kinesis streaming data sources.
Question 89
A company maintains an Amazon Redshift provisioned cluster that the company uses for extract, transform, and load (ETL) operations to support critical analysis tasks. A sales team within the company maintains a Redshift cluster that the sales team uses for business intelligence (BI) tasks. The sales team recently requested access to the data that is in the ETL Redshift cluster so the team can perform weekly summary analysis tasks. The sales team needs to join data from the ETL cluster with data that is in the sales team's BI cluster. The company needs a solution that will share the ETL cluster data with the sales team without interrupting the critical analysis tasks. The solution must minimize usage of the computing resources of the ETL cluster. Which solution will meet these requirements?
Show answer and explanation
Correct answer: A. Set up the sales team BI cluster as a consumer of the ETL cluster by using Redshift data sharing.
Redshift data sharing allows the sales team BI cluster to access data in the ETL cluster without duplicating data or impacting ETL performance. Data sharing is zero-copy, meaning queries against shared data do not consume compute resources from the source ETL cluster. The sales team can join their local data with shared data seamlessly. This meets all requirements: no interruption to critical ETL tasks, minimal resource consumption from the ETL cluster, and clean data access for the sales team.
Why the other options are wrong
- B. Direct access via materialized views still consumes ETL cluster resources and risks interruption to critical tasks.
- C. Direct access via database views has the same issues; it doesn't isolate the ETL cluster from additional query load.
- D. Weekly unload to S3 with Spectrum introduces staleness and extra complexity; data sharing provides real-time access without overhead.
Question 90
A company extracts approximately 1 TB of data every day from data sources such as SAP HANA, Microsoft SQL Server, MongoDB, Apache Kafka, and Amazon DynamoDB. Some of the data sources have undefined data schemas or data schemas that change. A data engineer must implement a solution that can detect the schema for these data sources. The solution must extract, transform, and load the data to an Amazon S3 bucket. The company has a service level agreement (SLA) to load the data into the S3 bucket within 15 minutes of data creation. Which solution will meet these requirements with the LEAST operational overhead?
Show answer and explanation
Correct answer: B. Use AWS Glue to detect the schema and to extract, transform, and load the data into the S3 bucket. Create a pipeline in Apache Spark.
AWS Glue is purpose-built for ETL operations with automatic schema detection capabilities through its crawler functionality. It handles multiple data sources (SAP HANA, SQL Server, MongoDB, Kafka, DynamoDB), manages undefined and evolving schemas, and can load data to S3 within the 15-minute SLA. Glue provides the least operational overhead compared to managing EMR clusters or Lambda functions with limited resources.
Why the other options are wrong
- A. Amazon EMR requires more operational management and cluster configuration compared to the managed Glue service.
- C. AWS Lambda has execution time limits and resource constraints that make it unsuitable for processing 1 TB daily within 15 minutes.
- D. Amazon Redshift with Spectrum is not designed for schema detection or ETL ingestion from diverse sources like Kafka and DynamoDB.
Question 91
A company uses a data lake that is based on an Amazon S3 bucket. To comply with regulations, the company must apply two layers of server-side encryption to files that are uploaded to the S3 bucket. The company wants to use an AWS Lambda function to apply the necessary encryption. Which solution will meet these requirements?
Show answer and explanation
Correct answer: B. Use dual-layer server-side encryption with AWS KMS keys (DSSE-KMS).
DSSE-KMS (Dual-layer Server-Side Encryption with KMS) provides two distinct layers of encryption for S3 objects. This is the only AWS-native solution that applies two separate layers of server-side encryption automatically without requiring manual implementation in Lambda. Option A combines SSE-KMS with the S3 Encryption Client but requires application-level implementation. Option C (SSE-C) requires customer-managed keys and must be applied before upload. Option D (SSE-KMS) provides only single-layer encryption.
Why the other options are wrong
- A. Requires manual application-level encryption implementation rather than native S3 feature
- C. SSE-C must be applied at upload time before Lambda can process, and provides only single-layer encryption
- D. Provides only single-layer encryption, not the required two layers
Question 92
A company uses Amazon RDS to store transactional data. The company runs an RDS DB instance in a private subnet. A developer wrote an AWS Lambda function with default settings to insert, update, or delete data in the DB instance. The developer needs to give the Lambda function the ability to connect to the DB instance privately without using the public internet. Which combination of steps will meet this requirement with the LEAST operational overhead? (Choose two.)
Show answer and explanation
Correct answer: C, D. Configure the Lambda function to run in the same subnet that the DB instance uses; Attach the same security group to the Lambda function and the DB instance. Include a self-referencing rule that allows access through the database port.
Configuring the Lambda function to run in the same subnet as the RDS instance allows VPC-based private communication. Attaching the same security group to both resources with a self-referencing ingress rule permits the Lambda function to connect to the database port privately. This approach requires no public access and leverages VPC security mechanisms for least operational overhead.
Why the other options are wrong
- A. Enabling public access exposes the database to the internet, violating the requirement for private connectivity.
- B. Allowing only Lambda invocations on the database port is not a valid security group rule specification.
- E. Network ACLs are stateless and don't automatically handle return traffic; security groups are the appropriate mechanism for this requirement.
That was 92 of 366.
The full AWS DEA-C01 pack has all 366 questions, each with the answer, the explanation and why the other options are wrong, plus a questions-only copy for timed runs. US$39, paid once, with free monthly updates and a pass-or-your-money-back guarantee.
Not ready yet? Get the first 183 free as one PDF by email. Get the Half Pack →