AWS Certified Data Analytics Test 1

Welcome to your AWS Certified Data Analytics - Test-1

Name

Phone

Q1. You are a data scientist working for a financial services company that has several relational databases, data warehouses, and NoSQL databases that hold transactional information about their financial trades and operational activities. The company wants to manage their financial counterparty risk through using their real-time trading/operational data to perform risk analysis and build risk management dashboards.

You need to build a data repository that combines all of these disparate data sources so that your company can perform their Business Intelligence (BI) analysis work on the complete picture of their risk exposure.
What collection system best fits this use case?

Select 1 option(s):

Financial data sources data -> Kinesis Data Firehose -> S3 -> Glue -> S3 Data Lake -> Athena

Financial data sources data -> Kinesis Data Firehose -> Kinesis Data Analytics -> Kinesis Data Firehose -> Redshift -> QuickSight

Financial data sources data -> Database Migration Service -> S3 -> Glue -> S3 Data Lake -> Athena

Financial data sources data -> Kinesis Data Streams -> Kinesis Data Analytics -> S3 Data Lake -> QuickSight

Q2. You are a data scientist working on a project where you have two large tables (orders and products) that you need to load into Redshift from one of your S3 buckets. Your table files, which are both several million rows large, are currently on an EBS volume of one of your EC2 instances in a directory titled $HOME/myredshiftdata.

Select 1 option(s):

Load your orders.tbl and products.tbl using the command: ‘aws s3 cp $HOME/myredshiftdata s3://dataanalytics/myredshiftdata --recursive’

Load your orders.tbl and products.tbl by first splitting each tbl file into smaller parts using the command: ‘split -d -l 5000000 -a 4 orders.tbl orders.tbl’ and ‘split -d -l 10000000 -a 4 products.tbl products.tbl’

Load your orders.tbl and products.tbl by first getting a count of the number of rows in each using the commands: ‘wc -l orders.tbl’ and ‘wc -l products.tbl’. Then splitting each tbl file into smaller parts using the command: ‘split -d -l # -a 4 orders.tbl orders.tbl’ and ‘split -d -l # -a 4 products.tbl products.tbl’ where # is replaced by the result of your wc command divided by 4.

Load your orders.tbl and products.tbl by first getting a count of the number of rows in each using the commands: ‘wc -l orders.tbl’ and ‘wc -l products.tbl’. Then splitting each tbl file into smaller parts using the command: ‘split -d -l # -a 4 orders.tbl orders.tbl-’ and ‘split -d -l # -a 4 products.tbl products.tbl-’ where # is replaced by the result of your wc command divided by 4.

Q3. You are working on a project where you need to perform real-time analytics on your application server logs. Your application is split across several EC2 servers in an auto-scaling group and is behind an application load balancer as depicted in this diagram:

You need to perform some transformation on the log data, such as joining rows of data, before you stream the data to your real-time dashboard.
What is the most efficient and performant solution to aggregate your application logs?

Select 1 option(s):

Install the Kinesis Agent on your application servers to watch your logs and use Kinesis Data Firehose to stream the logs directly to S3. Use Kinesis Data Analytics queries to build your real-time analytics dashboard.

Write a Kinesis Data Streams producer application that reads the application logs and pushes the data directly into your Kinesis data stream. Use Kinesis Data Analytics queries to build your real-time analytics dashboard.

Install the Kinesis Agent on your application servers to watch your logs and ingest the log data. Write a Kinesis Data Analytics application that reads the application log data from the agent, performs the required transformations, and pushes the data into your Kinesis data output stream. Use Kinesis Data Analytics queries to build your real-time analytics dashboard.

Use a CloudWatch dashboard that uses your application’s CloudWatch logs as the data source.

Q4. You are a data scientist on a team where you are responsible for ingesting IoT streamed data into a data lake for use in an EMR cluster. The data in the data lake will be used to allow your company to do business intelligence analytics on the IoT data. Due to the large amount of data being streamed to your application you will need to compress the data on the fly as you process it into your EMR cluster.

How would you most efficiently collect the data from your IoT devices?

Select 1 option(s):

Use the Kinesis REST API to get the IoT device records from the IoT devices and stream the data to your data lake through Kinesis Data Streams, then use Apache DistCp to move the data from S3 to your EMR cluster.

Use the AWS IoT service to get the device data from the IoT devices, use Kinesis Data Firehose to stream the data to your data lake, then use S3DistCp to move the data from S3 to your EMR cluster.

Use the Kinesis Producer Library to create a Kinesis producer application that reads the data from the IoT devices and stream the data to your data lake through Kinesis Data Streams, then use S3DistCp to move the data from S3 to your EMR cluster.

Use the Kinesis Client Library to get the device data from the IoT devices, use Kinesis Data Firehose to stream the data to your data lake, then use Apache DistCp to move the data from S3 to your EMR cluster.

Q5. You are a data scientist working for a rental car company that has fleets of rental cars across the globe. Each car is equipped with IoT sensors that report important information about the car’s functioning, location, service levels, mileage, etc.

You have been tasked with determining how rental efficiency has changed over time for fleets in certain cities across the US. This solution requires you to look back at several years of historical data collected by your IoT sensors.
Your management team wishes to perform Key Performance Indicator (KPI) analysis on the rental car data through visualization using business intelligence (BI) tools. They will use this analysis and visualization to make management decisions on how to keep their fleet of rental cars at optimum levels of service and use. They will also use the KPI analysis to assess the performance of their regional management teams for each city for which you collect data.
What collection system best fits this use case?

Select 1 option(s):

IoT device sensor data -> Kinesis Data Firehose -> S3 -> Glue -> S3 Data Lake -> Athena

IoT device sensor data -> Kinesis Data Firehose -> Kinesis Data Analytics -> Kinesis Data Firehose -> Redshift -> QuickSight

IoT device sensor data -> RDS -> Database Migration Service -> S3 -> Glue -> S3 Data Lake -> Athena

IoT device sensor data -> Kinesis Data Streams -> Kinesis Data Analytics -> S3 Data Lake -> QuickSight

Q6. You are a data scientist working for a mobile gaming company that is developing a new mobile gaming app that will need to handle thousands of messages per second arriving in your application data store. Due to the user interactivity of your game, all changes to the game datastore must be recorded with a before-change and after-change view of the data record. These data store changes will be used to deliver a near-real-time usage dashboard of the app for your management team.

What application collection system infrastructure best delivers these capabilities in the most performant and cost effective way?

Select 1 option(s):

Kinesis Firehose -> S3 -> EMR with Spark -> S3 -> Redshift -> QuickSight

DynamoDB -> DynamoDB Streams -> Lambda -> Kinesis Firehose -> Redshift -> QuickSight

Kinesis Firehose -> Aurora MySQL -> Lambda -> Kinesis Firehose -> Redshift -> QuickSight

Kinesis Data Streams -> Aurora MySQL -> Lambda->Kinesis Firehose -> Redshift -> QuickSight

Q7. You are a data scientist working for an online retail electronics chain. Their website receives very heavy traffic during certain months of the year, but these heavy traffic periods fluctuate over time. Your firm wants to get a better understanding of these patterns. Therefore, they have decided to build a traffic prediction machine learning model based on click-stream data.

Your task is to capture the click-stream data and store it in S3 for use as training and inference data in the machine learning model. You have built a streaming data capture system using Kinesis Data Streams and its Kinesis Producer Library (KPL) for your click-stream data capture component. You are using collection batching in your KPL code to improve performance of your collection system. Exception and failure handling is very important to your collection process, since losing click-stream data will compromise the integrity of your machine learning model data.
How can you best handle failures in your KPL component?

Select 1 option(s):

For each record processed by your KPL component trigger a Lambda function that ensures proper sequencing of the records processed

Kinesis Data Streams synchronously replicates your data across three availability zones. Take advantage of this to recover from failed record processing with retry logic.

With the KPL PutRecords operation, if a put fails, the record is automatically put back into the KPL buffer and retried.

With the KPL PutRecords operation, if a put fails, the record is automatically rolled back, giving you the option to use retry logic in your KPL code.

Q8. You are a data scientist working for a large city that has implemented an electric scooter ride sharing system. Each electric scooter is equipped with IoT sensors that report the scooter’s location, whether it is currently rented out, current renter, battery level, speed of travel, etc.

You have been tasked with determining scooter density of location throughout the city and redistributing scooters if some areas of the city are overpopulated with scooters while other areas are underpopulated. This solution requires real-time IoT data to be ingested into your data collection system.
Your management team wishes to perform real-time analysis on the scooter data through visualization using business intelligence (BI) tools. They will use this analysis and visualization to make management decisions on how to keep their fleet of scooters at optimum levels of service and use.
What collection system best fits this use case?

Select 1 option(s):

IoT device sensor data -> Kinesis Data Firehose -> S3 -> Glue -> S3 Data Lake -> Athena

IoT device sensor data -> Kinesis Data Firehose -> Kinesis Data Analytics -> Kinesis Data Firehose -> Redshift -> QuickSight

IoT device sensor data -> RDS -> Database Migration Service -> S3 -> Glue -> S3 Data Lake -> Athena

IoT device sensor data -> Kinesis Data Streams -> Kinesis Data Analytics -> S3 Data Lake -> QuickSight

Q9. You are a data scientist working for a medical services company that has a suite of apps available for patients and their doctors to share their medical data. These apps are used to share patient details, MRI and XRAY images, appointment schedules, etc. Because of the importance of this data and its inherent Personally Identifiable Information (PII), your data collection system needs to be secure and the system cannot suffer lost data, process data out of order, or duplicate data.

Which data collection system ( s ) gives you the security and data integrity your requirements demand? (SELECT 2)

Select 2 option(s):

Apache Kafka/Amazon MSK

SQS (FIFO)

SQS (Standard)

Kinesis Data Firehose

Kinesis Data Streams

DynamoDB Streams

Q10. You work for a ski resort corporation. Your company is developing a lift ticket system for mobile devices that allows skiers and snowboarders to use their phone as their lift ticket. The ski resort corporation owns many resorts around the world. The lift ticketing system needs to handle users who move from resort to resort throughout any given time period. Resort customers can also purchase packages where they can ski or snowboard at a defined list (a subset of the total) of several different resorts across the globe as part of their package.

The storage system for the lift ticket mobile application has to handle large fluctuations in volume. The data collected from the devices and stored in the data store is small in size, but the system must provide the data at low latency and high throughput. It also has to authenticate users through their mobile device registered facial recognition service, so that users can’t share a lift ticket by sharing their mobile devices.
What storage system is the best fit for this system?

Select 1 option(s):

Neptune

RDS

DynamoDB

ElastiCache

Redshift

Q11. You work for a mobile gaming company that has developed a word puzzle game that allows multiple users to challenge each other to complete a crossword puzzle type of game board. This interactive game works on mobile devices and web browsers. You have a world-wide user base that can play against each other no matter where each player is located.

You now need to create a leaderboard component of the game architecture where players can look at the daily point leaders for the day, week, or other timeframes. Each time a player accumulates points, the points counter for that player needs to be updated in real-time. This leaderboard data is transient in that it only needs to be stored for a limited duration.
Which of the following architectures best suits your data access and retrieval patterns using the simplest, most efficient approach?

Select 1 option(s):

Data sources -> Kinesis Data Streams -> Spark Streaming on EMR -> ElastiCache Redis -> DynamoDB

Data Sources -> Kinesis Data Firehose -> S3 -> Athena

Data sources -> Kinesis Data Streams -> Spark Streaming on EMR -> ElastiCache Memcached -> DynamoDB

Data sources -> Kinesis Data Streams -> Spark Streaming on EMR -> ElastiCache Redis

Data sources -> Kinesis Data Firehose -> Spark Streaming on EMR -> ElastiCache Redis -> S3

Q12. You work for a car manufacturer who has implemented many sensors into their vehicles such as GPS, lane-assist, braking-assist, temperature/humidity, etc. These cars continuously transmit their structured and unstructured sensor data. You need to build a data collection system to capture their data for use in ad-hoc analytics applications to understand the performance of the cars, the locations traveled to and from, the effectiveness of the lane and brake assist features, etc. You also need to filter and transform the sensor data depending on rules based on parameters such as temperature readings. The sensor data needs to be stored indefinitely, however you only wish to pay for the analytics processing when you use it.

Which of the following architectures best suits your data lifecycle and usage patterns using the simplest, most efficient approach?

Select 1 option(s):

Sensor data -> Kinesis Data Streams -> IoT Core -> S3 -> Athena

Sensor data -> Kinesis Data Firehose -> IoT Core -> S3 -> Athena

Sensor data -> Kinesis Data Streams -> IoT Core -> Kinesis Data Firehose -> RedShift -> QuickSight

Sensor data -> IoT Core -> S3 -> Athena

Sensor data -> Kinesis Data Firehose -> S3 -> Athena

Q13. You work for a public health governmental organization where you are responsible for building out a data warehouse to hold infectious disease information based on the data found at the World Health Organization’s Global Health Observatory data repository. You expect your initial data warehouse to hold less than TBs of data. However, you expect that the data stored in your warehouse will grow rapidly based on the state of world-wide infectious disease progression in the near future.

Your organization plans to use the data stored in your data warehouse to visualize disease progression across the various states in your country as infectious diseases progress through their lifecycle. These analyses will be used to make important decisions about citizen interaction and mobility.
Which of the following data warehouse configurations best suits your data analysis scenario using the simplest, most cost effective approach?

Select 1 option(s):

Redshift with RA3 nodes

Redshift with DC2 nodes

S3 with SSD volumes

S3 with HDD volumes

Redshift with DS2 nodes

Q14. You work for a large city police department as a data scientist. You have been given the task of tracking crime by city district for each criminal committing the given crime. You have created a DynamoDB table to track the crimes across your city’s districts. The table has this configuration: for each crime the table contains a CriminalId (the partition key), CityDistrict, and CrimeDate the crime was reported. Your police department wants to create a dashboard of the crimes reported by district and date.

What is the most cost effective way to retrieve the crime data from your DynamoDB table to build your crimes reported by district and date?

Select 1 option(s):

Create a local secondary index with CriminalId as the partition key and CrimeDate as the sort key

Create a global secondary index with CityDistrict as the partition key and CrimeDate as the sort key

Scan the table and use the ProjectionExpression parameter to return the crimes reported by district and date

Scan the secondary index and use the ProjectionExpression parameter to return the crimes reported by district and date

Q15. You work for a large retail and wholesale business with a significant ecommerce web presence. Your company has just acquired a new ecommerce clothing line and needs to build a data warehouse for this new line of business. The acquired ecommerce business sells clothing to a niche market of men’s casual and business attire. You have chosen to use Amazon Redshift for your data warehouse. The data that you’ll initially load into the warehouse will be relatively small. However, you expect the warehouse data to grow as the niche customer base expands once the parent company makes a significant investment in advertising.

What is the most cost effective and best performing Redshift strategy that you should use when you create your initial tables in Redshift?

Select 1 option(s):

Use the KEY distribution strategy

Use the EVEN distribution strategy

Use the ALL distribution strategy

Use the AUTO distribution strategy

Q16. You are a data scientist working for a multinational conglomerate corporation that has many data stores for which you need to provide a common repository. All of your company’s systems need to use this common repository to store and retrieve metadata to work with the data stored in all of the data siolos throughout the organization. You also need to provide the ability to query and transform the data in the organization’s data silos. This common repository will be used for data analytics by your data scientist team to produce dashboards and KPIs for your management team.

You are using AWS Glue to build your common repository as depicted in this diagram:

As you begin to create this common repository you notice that you aren’t getting the inferred schema for some of your data stores. You have run your crawler against your data stores using your custom classifiers. What might be the problem with your process?

Select 1 option(s):

The username you provided to your JDBC connection to your S3 buckets does not have SELECT permission to retrieve metadata from the S3 bucket data store

The username you provided to your JDBC connection to your Redshift clusters does not have SELECT permission to retrieve metadata from the Redshift data store

You did not use the Glue built-in classifiers in your crawler job

The username you provided to your JDBC connection to your DynamoDB tables does not have SELECT permission to retrieve metadata from the DynamoDB data store

Q17. You are a data scientist working for a retail chain that stores information about their supply chain partners (partner metadata) and their interaction with these partners (products produced, payments processed, competing partners, etc.). You are tasked with building a data store and associated data lifecycle management system for this partner data. The data will be used for analytics in managing these partners to maximize profitability for your supply chain.

You need to manage the data lifecycle according to the various access patterns defined for each type while maintaining storage cost efficiency. The partner metadata is less frequently accessed than the partner interaction data. You need to manage your storage costs so that high frequency accessed data (such as your partner interaction data) is available at very fast response times (sub-second), less frequently accessed data (such as your partner metadata) is available in minutes, and your rarely accessed data (such as historical data on former partners) is available within hours.

Which storage lifecycle best fits your usage patterns and business requirements?

Select 1 option(s):

Partner interaction data (sub-second response) stored in Redshift, partner metadata (minutes response) stored in S3 Standard, and former partner data (hours response) in S3 Intelligent-Tiering.

Use a Redshift cluster for all of your data. Create RA3 nodes in your cluster for your partner interaction data (sub-second response), create DC2 nodes for your partner metadata (minutes response), and DS2 nodes for your former partner data (hours response).

Partner interaction data (sub-second response) stored in Redshift, partner metadata (minutes response) stored in S3 Standard, and former partner data (hours response) in S3 Glacier.

Partner interaction data (sub-second response) stored in RDS Aurora, partner metadata (minutes response) stored in S3 Standard, and former partner data (hours response) in S3 Glacier.

Q18. You are a data analyst working for a scientific research and data science company that is building a large scale data lake on EMR to house research data for ongoing research projects. Some of the projects have data processing requirements that need hot data set access, while others require less-hot data set access. For example, analysis for political polling related projects requires hot data set access due to the pressing nature of understanding political analytics and trends in real-time. Infrastructure and materials projects have less-hot data set access requirements since these projects have the option of producing their analysis on a daily basis versus a real-time basis.

Additionally, the real-time analytics projects require fast performance, their data is considered timely but temporary. However, the less-hot data projects don’t require real-time analytics, they require persistent data storage.

Which data processing solution best fits your usage patterns and business requirements?

Select 1 option(s):

S3 BFS for the hot data sets, S3 Glacier for the less-hot data sets

S3 EMRFS for the hot data sets, HDFS for the less-hot data sets

HDFS for the hot data sets, S3 EMRFS for the less-hot data sets

S3 BFS for the hot data sets, HDFS for the less-hot data sets

Q19. You are a data scientist working for a large transportation company that manages its distribution data across all of its distribution lines: trucking, shipping, airfreight, etc. This data is stored in a data warehouse in Redshift. The company ingests all of the distribution data into an EMR cluster before loading the data into their data warehouse in Redshift. The data is loaded from EMR to Redshift on a schedule, once per day.

How might you lower the operational costs of running your EMR cluster? (Select TWO)

Select 2 option(s):

EMR Transient Cluster

EMR Long-running Cluster

EMR Core Nodes as spot instances

EMR Task Nodes as spot instances

EMR cluster launched via AWS CLI using defaults

Q20. You are a data scientist working for an online retail company that wishes to catalog all of their products in a data lake. They also want to load their product data from their data lake into a data warehouse that they can use for business intelligence (BI) dashboards and analytics with QuickSight.

How would you automate and operationalize the data processing to get the company’s product data from their data lake to their data warehouse in the most efficient, cost effective manner?

Select 1 option(s):

Product data put to S3 data lake -triggers-> Lambda -runs-> Glue Crawler -> on completion CloudWatch event rule -triggers-> Lambda which runs Glue ETL job that transforms data to JSON -> S3 -triggers-> Lambda which runs COPY command to move data to Redshift

Product data put to S3 data lake -triggers-> Lambda -runs-> Glue Crawler -> on completion CloudWatch event rule -triggers-> Lambda which runs Glue ETL job that transforms data to Parquet -> S3 -triggers-> Lambda which runs COPY command to move data to Redshift

Product data put to S3 data lake -triggers-> Lambda -runs-> Glue Crawler -> on completion CloudWatch event rule -triggers-> Lambda which runs Glue ETL job that transforms data to CSV -> S3 -triggers-> Lambda which runs COPY command to move data to Redshift

Q21. You work as a data scientist at a large hedge fund. Your firm produces analytics dashboard data for all of its traders. The data that you use is extracted from several trading systems, then transformed by removing canceled trades and classifying trades that remain open as pending. Quite often there are exotic trade types that your analytics application has not processed in past runs. When this happens your data processing solution needs to handle these new types of trades without having to modify the transformation code or the downstream data store.

This process is run at the end of each trading day for each trader in the firm. How would you automate and operationalize this data processing flow in the most efficient, cost effective manner?

Select 1 option(s):

A Glue schedule trigger runs at the end of the day which starts two Glue transformation jobs: remove_canceled_trades and classify_open_trades_as_pending. When both of these jobs have completed an event trigger starts a Glue crawler that crawls the transformed data and updates the schema. Upon completion of the crawler schema update, a Glue ETL job runs and uses the COPY command to move the data to Redshift. Analytics dashboards are built using Redshift data.

A Glue schedule trigger runs at the end of the day which starts two Glue transformation jobs: remove_canceled_trades and classify_open_trades_as_pending. When both of these jobs have completed an event trigger starts a Glue crawler that crawls the transformed data and updates the schema. Upon completion of the crawler schema update, a Glue ETL job runs and uses the UNLOAD command to move the data to Redshift. Analytics dashboards are built using Redshift data.

A cron job schedule trigger runs at the end of the day which starts two Glue transformation jobs: remove_canceled_trades and classify_open_trades_as_pending. When both of these jobs have completed a cron job schedule trigger starts a Glue crawler that crawls the transformed data and updates the schema. Upon completion of the crawler schema update, a Glue ETL job runs and uses the COPY command to move the data to Redshift. Analytics dashboards are built using Redshift data.

A Glue schedule trigger runs at the end of the day which starts two Glue transformation jobs: remove_canceled_trades and classify_open_trades_as_pending. When both of these jobs have completed an event trigger starts a Glue crawler that crawls the transformed data and updates the schema. Upon completion of the crawler schema update, a Glue ETL job runs and uses the PUT command to move the data to Redshift. Analytics dashboards are built using Redshift data.

Q22. You work as a data scientist at a large global bank. Your bank receives loan information in the form of weekly files from several different loan processing and credit verification agencies. You need to automate and operationalize a data processing solution to take these weekly files, transform them and then finish up by combining them into one file to be ingested into your Redshift data warehouse. The files arrive at different times every week, but the delivering agencies attempt to meet their service level agreement (SLA) of 1:00 AM to 4:00 AM. Unfortunately, the agencies frequently miss their SLAs. You have a tight batch time frame into which you have to squeeze all of this processing.

How would you build a data processing system that allows you to gather the agency files and process them for your data warehouse in the most efficient manner and in the shortest time frame?

Select 1 option(s):

Agency files arrive on an S3 bucket. An ETL Lambda function is triggered as each file arrives. The ETL Lambda function transforms the data and writes the transformed file to another S3 bucket. After all of the agency files have been processed by the ETL Lambda function, another Lambda function is triggered to combine the agency file data into one parquet file and write it to another S3 bucket. Then a last Lambda function is triggered to run the COPY command to load the parquet file data into Redshift.

Agency files arrive on an S3 bucket. Use CloudWatch events to schedule a weekly Step Functions state machine. The Step Functions state machine calls a Lambda function to verify that the agency files have arrived. The state machine then starts several Glue ETL jobs in parallel to transform the agency data. Once the agency file transformation jobs have completed the state machine starts another Glue ETL job to combine the transformed agency files and convert the data to a parquet file. The parquet file is written to an S3 bucket. Then the state machine finally runs a last Glue ETL job to run the COPY command to load the parquet file data into Redshift.

Agency files arrive on an S3 bucket. An ETL Lambda function is triggered as each file arrives. The ETL Lambda function transforms the data and writes the transformed file to another S3 bucket. After all of the agency files have been processed by the ETL Lambda function, another Lambda function is triggered to combine the agency file data into one CSV file and write it to another S3 bucket. Then a last Lambda function is triggered to run the UNLOAD command to load the CSV file data into Redshift.

Agency files arrive on an S3 bucket. Use CloudWatch events to schedule a weekly Step Functions state machine. The Step Functions state machine calls a Lambda function to verify that the agency files have arrived. The state machine then starts several Glue ETL jobs in parallel to transform the agency data. Once the agency file transformation jobs have completed the state machine starts another Glue ETL job to combine the transformed agency files and convert the data to an ORC file. The ORC file is written to an S3 bucket. Then the state machine finally runs a last Glue ETL job to run the UNLOAD command to load the ORC file data into Redshift.

Q23. You work as a cloud architect for a cloud consultancy practice at a major IT consulting firm. Your latest client has a series of data processing Apache Spark ELT jobs that they want to run in a pipeline on EMR. Thay have asked you which set of data processing tools and techniques will best suit their pipeline needs. The jobs have a specified sequence. Your client wants to manage their costs. Therefore, they want to keep the solution simple, they don’t want to build an application to run these jobs, and they don’t want to incur any additional costs on virtual servers to run their pipeline. Also, they plan on integrating their Apache Spark pipeline with other AWS services in the future.

Which orchestration tool set best suits your client’s pipeline requirements?

Select 1 option(s):

Apache Oozie to schedule and run the Spark jobs

Apache Airflow to schedule and run the Spark jobs

AWS Step Functions to schedule and run the Spark jobs

AWS Lambda to schedule and run the Spark jobs

AWS DMS to schedule and run the Spark jobs

Q24. You work as a cloud architect for a gaming company that is building an analytics platform for their gaming data. This analytics platform will ingest game data from current games being played by users of their mobile game platform. The game data needs to be loaded into a data lake where business intelligence (BI) tools will be used to build analytics views of key performance indicators (KPIs). You load your data lake from an EMR cluster where you run Glue ETL jobs to perform the transformation of the incoming game data to the parquet file format. Once transformed, the parquet files are stored in your S3 data lake. From there you can run BI tools, such as Athena, to build your KPIs.

You want to handle EMR step through recovery logic. What is the simplest way to build retry logic into your data processing solution?

Select 1 option(s):

CloudTrail event rule sends a text message via a Simple Notification Service (SNS) topic, a support engineer reruns the failed EMR step.

CloudWatch event rule sends a text message via a Simple Notification Service (SNS) topic, a support engineer reruns the failed EMR step.

CloudTrail event rule triggers a Lambda function via a Simple Notification Service (SNS) topic which retries the EMR step.

CloudWatch event rule triggers a Lambda function via a Simple Notification Service (SNS) topic which retries the EMR step.

CloudWatch event rule triggers a retry of the Spark step via a Simple Notification Service (SNS) topic.

Q25. You work as a cloud security architect for a financial services company. Your company has an EMR cluster that is integrated with their AWS Lake Formation managed data lake. You use the Lake Formation service to enforce column-level access control driven by policies you have defined. You need to implement a real-time alert and notification system if authenticated users run the TerminateJobFlows, DeleteSecurityConfiguration, or CancelSteps actions within EMR.

How would you implement this real-time alert mechanism in the simplest way possible?

Select 1 option(s):

Create a CloudTrail trail and enable continuous delivery of events to an S3 bucket. Use the aws cloudtrail create-trail CLI command to create an SNS topic. When an event occurs a Simple Queue Service (SQS) queue that subscribes to the SNS topic will receive the message. Use a Lambda function triggered by SQS to filter the messages for the TerminateJobFlows, DeleteSecurityConfiguration, or CancelSteps actions. The Lambda function will notify security alert subscribers via another SNS topic.

Create a CloudWatch event and enable continuous delivery of events to an S3 bucket. Use the aws cloudwatch create-event CLI command to create an SNS topic. When an event occurs for the TerminateJobFlows, DeleteSecurityConfiguration, or CancelSteps actions subscribers to the SNS topic will be notified.

Create a Lambda function that subscribes to an SNS topic that you define. The Lambda function will be triggered every time a TerminateJobFlows, DeleteSecurityConfiguration, or CancelSteps action is written to the EMR logs.

Create a CloudTrail trail and enable continuous delivery of events to an S3 bucket. Use the aws cloudtrail create-trail CLI command to create an SNS topic. When an event occurs for the TerminateJobFlows, DeleteSecurityConfiguration, or CancelSteps actions SNS will notify security alert subscribers.

Q26. You work as a data scientist for a medical data processing company. Your company receives patient data via file feeds into one of your S3 buckets. The data is formatted as a nested JSON document similar to this:

[

{

“id”: “796”,

“category”: “Epidemiology”,

“info”: {

“subcategory”: “Neuroepidemiology”,

“questionType”: “multiple choice 1”,

“question”: “What is the reference to pi?”,

“answers”: [

“First three digits”,

“Infinite number of digits”,

“Digits after the decimal point”,

“Digits before the decimal point”

“correctAnswer”: [“Digits after the decimal point”]

}

]

After performing data engineering on some sample files you have noticed occasional inconsistencies in the data types in the JSON.
What is the most performant and cost effective way to clean your semi-structured JSON data?

Select 1 option(s):

Run an AWS Batch job that uses the dirtyjson library

Use an EMR job that uses the Spark native DataFrame API

Trigger a Lambda function that uses the json.load and json.loads libraries

Run a Glue job that uses the DynamicFrame extension

Q27. You work as a data scientist for a rideshare company. Rideshare request data is collected in one of the company’s S3 buckets (inbound bucket). This data needs to be processed (transformed) very quickly, within seconds of being put onto the S3 bucket. Once transformed, the rideshare request data must be put into another S3 bucket (transformed bucket) where it will be processed to link rideshare drivers with rideshare requesters.

You have already written Spark jobs to do the transformation. You need to control costs and minimize data latency for the rideshare request transformation operationalization of your data collection system. Which option best meets your requirements?

Select 1 option(s):

Lambda function triggered when the rideshare data request is put onto the inbound S3 bucket. Lambda sends an SNS topic to an SQS queue. Another Lambda function polls the queue every minute and when it finds a message it launches an EMR cluster and submits a Spark job to process the request.

Lambda function triggered when the rideshare data request is put onto the inbound S3 bucket. The Lambda function passes the request data to a Spark job in Glue.

Lambda function triggered when the rideshare data request is put onto the inbound S3 bucket. The Lambda function launches an EMR cluster and submits the job using the EMR Steps API to process the request.

Build an EMR cluster that runs Apache Livy. Lambda function triggered when the rideshare data request is put onto the inbound S3 bucket. The Lambda function passes the request data to a Spark job on the EMR cluster.

Q28. You are a data scientist working on a preventative health research project using the Global Health Observatory data repository. This repository contains the Body Mass Index (BMI) dataset which is based on several thousand observations from around the globe from 1975 to 2016. You need to analyze this dataset using QuickSite. One of the visuals you’ve been asked to create is to show the prevalence of thinness by country across the globe from 1975 to 2016 at 5-year increments. What is the best visual type to use to display this data?

Select 1 option(s):

Geospatial chart

Bubble chart

Heat map

Tree map

Q29. You are a data scientist working for the Fédération Internationale de Football Association (FIFA). Your management team has asked you to select the appropriate data analysis solution to analyze streaming football data in near real-time. You need to use this data to build interactive results through graphics and interactive charts for the FIFA management team. The football streaming events are based on time series that are unordered and may frequently be duplicated. You also need to transform the football data before you store it. You’ve been instructed to focus on providing high quality functionality based on fast data access.

Which solution best fits your needs?

Select 1 option(s):

Kinesis data firehose -> ORC files -> S3 -> Athena

Kinesis data firehose -> Lambda -> Elasticsearch Cluster -> Kibana

Kinesis data firehose -> RDS -> QuickSight

Kinesis Data Streams -> parquet files -> S3 -> Redshift Spectrum

Q30. You are a data scientist working for a large bank where you are building out an EMR cluster for their customer information data lake. Due to the Personally Identifiable Information (PII) stored in the data lake, you need to lock down all environments (dev, engineering, test, perf, prod) to make sure only the appropriate users and user groups have access to the data lake.

To accomplish this goal you have created this IAM policy and attached it to your users and user groups who will be working with your EMR cluster:
{
“Version”: “2012-10-17”,
“Statement”: [
{
“Sid”: “Stmt7645587658758”,
“Effect”: “Allow”,
“Action”: [
“elasticmapreduce:DescribeCluster”,
“elasticmapreduce:ListSecurityConfigurations”,
“elasticmapreduce:ListSteps”,
“elasticmapreduce:TerminateJobFlows”,
“elasticmapreduce:ModifyCluster”,
“elasticmapreduce:PutAutoScalingPolicy”,
“elasticmapreduce:ListInstances”,
“elasticmapreduce:SetTerminationProtection”,
“elasticmapreduce:DescribeStep”
],
“Resource”: [
“*”
],
“Condition”: {
“StringEquals”: {
“elasticmapreduce:ResourceTag/department”: [“dev”, “eng”]
}
}
}
]
}
How does this policy protect your EMR cluster that contains the company’s customer PII data?

Select 1 option(s):

It prevents users from performing the actions listed in the Action part of the Statement

It allows users to perform only the actions listed in the Action part of the Statement

It only allows users to perform the actions listed in the Action part of the Statement if the EMR cluster they are attempting to access is tagged as department of dev or eng

It only allows users to perform the actions listed in the Action part of the Statement if the EMR cluster they are attempting to access is NOT tagged as department of dev or eng

Q31. You are a data scientist working for a financial services firm where you are building out an EMR cluster used to house the data lake used for your company’s proprietary machine learning models that predict market movement in global markets. The data in this data lake is considered to be a fundamental part of the company’s knowledge capital so it can only be accessed by users in the Quantitative Equity Group defined within IAM. You need to lock down all environments (dev, engineering, test, perf, prod) to make sure only the users in the Quantitative Equity Group have access to the data lake.

To accomplish this goal you have created this IAM policy and attached it to your users in the Quantitative Equity Group IAM group who will be working with your confidential EMR cluster:
{
“Version”: “2012-10-17”,
“Statement”: [
{
“Sid”: “Stmt7645587658758”,
“Effect”: “Allow”,
“Action”: [
“elasticmapreduce:DescribeCluster”,
“elasticmapreduce:ListSecurityConfigurations”,
“elasticmapreduce:ListSteps”,
“elasticmapreduce:TerminateJobFlows”,
“elasticmapreduce:ModifyCluster”,
“elasticmapreduce:PutAutoScalingPolicy”,
“elasticmapreduce:ListInstances”,
“elasticmapreduce:SetTerminationProtection”,
“elasticmapreduce:DescribeStep”
],
“Resource”: [
“*”
],
“Condition”: {
“StringEquals”: {
“elasticmapreduce:ResourceTag/department”:
[“dev”, “eng”, “test”, “perf”, “prod”]
}
}
}
]
}

You then created this policy and attached it to all users to further lockdown the EMR cluster environments:

{
“Version”: “2012-10-17”,
“Statement”: [
{
“Effect”: “Deny”,
“Action”: [
“elasticmapreduce:AddTags”,
“elasticmapreduce:RemoveTags”
],
“Condition”: {
“StringNotEquals”: {
“elasticmapreduce:ResourceTag/department”:
[“dev”, “eng”, “test”, “perf”, “prod”]
}
},
“Resource”: [
“*”
]
}
]
}
What further protection does this policy give you (SELECT TWO)?

Select 2 option(s):

It prevents users from adding or removing tags on an EMR cluster that is tagged with the department of dev, eng, test, perf, or prod. This prevents users from giving themselves access to the EMR cluster by adding a tag to the cluster that their user profile allows

It prevents users from adding or removing tags on an EMR cluster that is tagged with the department of dev, eng, test, perf, or prod. This prevents users from giving anyone access to the EMR cluster by adding tags to the cluster, opening up access to any user

It prevents users from adding or removing tags on an EMR cluster that is tagged with the department of dev, eng, test, perf, or prod. This prevents users from giving themselves access to the EMR cluster by removing the tags on the cluster, opening up access to any user even if they aren’t in the Quantitative Equity Group IAM group

It prevents users from adding or removing tags on an EMR cluster that is tagged with the department of dev, eng, test, perf, or prod. This prevents users from giving themselves access to the EMR cluster by removing a tag on the cluster, opening up access to any user who is allowed to act on resources with that tag

It prevents users from adding or removing tags on an EMR cluster that is tagged with the department of dev, eng, test, perf, or prod. This prevents users from creating a new EMR cluster by cloning the Quantitative Equity Group owned cluster into the new cluster

Q32. You are a data scientist working for a medical services firm where you are building out an EMR cluster used to house the data lake used for your company’s client healthcare protected health information (PHI) data. The storage of this type of data is highly regulated through the Health Insurance Portability and Accountability Act (HIPAA). Specifically, HIPAA requires that healthcare companies, like your company, encrypt their client’s PHI data using encryption technology.

You have set up your EMR cluster to use the default of using the EMRFS to read and write your client’s PHI data to and from S3. You need to encrypt your client’s PHI data before you send it to S3.
Which option is the best encryption technique to use for your EMR cluster configuration?

Select 1 option(s):

SSE-S3

SSE-KMS

CSE-KMS

SSE-C

Q33. You are a data scientist working for a retail clothing manufacturer that has a large onlinepresence through their retail website. The website gathers Personally Identifiable Information (PII), such as credit card numbers, when customers complete their purchases on the website. Therefore, your company must adhere to the Payment Card Industry Data Security Standard (PCI DSS). Your company wishes to store the client data and purchase information data gathered through these transactions in their data warehouse, running on Redshift, where they intend to build Key Performance Indicator (KPI) dashboards using QuickSight.

You and your security department know that your data collection system needs to obfuscate the PII (credit card) data, gathered through your data collection system. How should you protect the highly sensitive credit card data in order to meet the PCI DSS requirements while keeping your data collection system as efficient and cost effective as possible?

Select 1 option(s):

AWS Shield Advanced for website traffic

AWS WAF for website traffic

Tokenization PII data

Use GuardDuty for website traffic

KMS encryption in transit and at rest

Q34. You are a data scientist working for a sports gambling company that produces sports betting data for inclusion in mainstream sports websites. Your company’s data is proprietary and needs to be protected for copyright purposes. You have been tasked with creating a data lake on S3 and also loading a relational database that stores your sports data. Any parameters (such as database connection information) used when building analytics applications used to access the data lake and/or database need to be stored in a secure service that encrypts the parameters. Your management team also has the requirement that parameters like database connection information be rotated automatically.

What AWS service should you use to protect the media content and metadata?

Select 1 option(s):

AWS IAM

AWS Systems Manager Parameter Store

KMS encryption

AWS Secrets Manager

Q35. You are a data scientist working for a healthcare company that needs to comply with Health Insurance Portability and Accountability Act (HIPAA) regulations. Your company needs to take all of their patient’s data, including test diagnostic data, wearable sensor data, diagnostic data from all doctor visits, etc. and store it in a data lake. They then want to use Athena and other Business Intelligence (BI) tools to query the patient data to enable their healthcare providers to give optimal service to their patients.

In order to apply the appropriate data governance and compliance controls, what AWS service(s) will allow you to provide the appropriate (HIPAA) reports? Also, what AWS service(s) will allow you to monitor changes to your data lake S3 bucket ACLs and bucket policies to scan for public read/write access violations?

Select 1 option(s):

CloudTrail to gather the data for the Business Associate Addendum (BAA) HIPAA compliance report. Use custom rules in AWS Config to track and report on S3 ACL and/or bucket policy changes that violate your security policies.

CloudWatch to gather the data for the Business Associate Addendum (BAA) HIPAA compliance report. Use custom rules in AWS Resource Access Manager to track and report on S3 ACL and/or bucket policy changes that violate your security policies.

AWS Artifact to retrieve the Business Associate Addendum (BAA) HIPAA compliance report. Use custom rules in AWS Config to track and report on S3 ACL and/or bucket policy changes that violate your security policies.

AWS Artifact to retrieve the Business Associate Addendum (BAA) HIPAA compliance report. Use custom rules in AWS Resource Access Manager to track and report on S3 ACL and/or bucket policy changes that violate your security policies.

Q36. You are a data scientist working for a company that provides credit card verification services to banks and insurance companies. Your client credit card data is streamed into your S3 data lake on a daily basis in the form of large sets of JSON files. Due to the Personally Identifiable Information (PII) data contained in these JSON files, your company must adhere to the regulations defined in the Payment Card Industry Data Security Standard (PCI DSS). This means you must encrypt the data at rest in your S3 buckets. You also need to recognize and take action on any abnormal data access activity.

Which option best satisfies your data governance and compliance controls in the most cost effective manner?

Select 1 option(s):

Store the credit card JSON data in DynamoDB with encryption enabled on your tables. Use a Lambda function to determine if any of your compliance rules are violated by scanning the DynamoDB tables. When compliance rule violations are found, send alerts using Simple Notification Service (SNS).

Store the credit card JSON data in buckets in S3 with encryption enabled. Use a Lambda function to determine if any of your compliance rules are violated by scanning the S3 buckets. When compliance rule violations are found, send alerts using Simple Notification Service (SNS).

Store the credit card JSON data in buckets in S3 with encryption enabled. Use the AWS Macie service to determine if any of your compliance rules are violated by scanning the S3 buckets. When compliance rule violations are found, use CloudWatch events to trigger alerts sent via Simple Notification Service (SNS).

Store the credit card JSON data in DynamoDB with encryption enabled on your tables. Use the AWS Macie service to determine if any of your compliance rules are violated by scanning the DynamoDB tables. When compliance rule violations are found, use CloudWatch events to trigger alerts sent via Simple Notification Service (SNS).

Q37. You are a data scientist working for a company that processes industrial machine operational data for various industrial manufacturers around the globe. You receive streaming data via Kinesis Data Firehose from the various manufacturers. You want to ingest the data into your Splunk cluster to deliver operational intelligence analysis, security analytics, and business performance KPIs for your manufacturing clients.

You have installed your Splunk cluster within your VPC. However, you have noticed that the ingestion process of moving your data from Kinesis Data Firehose to your Splunk cluster is failing. Which configuration option will allow your Kinesis Data Firehose stream to move your data into your Splunk cluster?

Select 1 option(s):

Change the Splunk cluster security group to allow access from Kinesis service IPs

Change your S3 bucket policy to allow Kinesis Data Firehose to write to your S3 bucket to allow access from service IPs

Verify the IAM role assigned to your Kinesis Data Firehose stream allows access from service IPs

Change your Splunk cluster ACL to allow access from Kinesis service IPs

Q38. You are a data scientist working for a large hedge fund. Your hedge fund managers rely on analytics data produced from the S3 data lake you have built that houses trade data produced by the firm’s various traders. You are configuring a public Elasticsearch domain that will allow your hedge fund managers to gain access to your trade data stored in your data lake. You have given your hedge fund managers Kibana to allow them to use visualizations you’ve produced to manage their traders activity.

When your hedge fund managers first test out your Kibana analytics visualizations, you find that Kibana cannot connect to your Elasticsearch cluster. Which options are ways to securely give your hedge fund managers access to your Elasticsearch cluster via their Kibana running on their local desktop? (SELECT TWO)

Select 2 option(s):

Configure a proxy server that acts as an intermediary between your Kibana users and your Elasticsearch cluster. Add an IP-based access IAM policy which allows requests from your user’s IP address to gain access to your Elasticsearch cluster through the proxy server’s IP address.

Configure a proxy server that acts as an intermediary between your Kibana users and your Elasticsearch cluster. Configure an open access IAM policy which allows requests from your user’s IP address to gain access to your Elasticsearch cluster through the proxy server’s IP address.

Configure a proxy server that acts as an intermediary between your Kibana users and your Elasticsearch cluster. Configure a security group which allows requests from your user’s IP address to gain access to your Elasticsearch cluster through the proxy server’s IP address.

Setup Amazon Cognito by creating a user pool and an identity pool to authenticate your Kibana users

Add an IP-based access IAM policy and use a security group which allows requests from your user’s IP address to gain access to your Elasticsearch cluster through the proxy server’s IP address.

Q39. You have just landed a new job as a data scientist for a worldwide retail and wholesale business with distribution centers located all around the globe.Your first assignment is to build a data collection system that stores all of the company’s product distribution performance data from all of their distribution centers into S3. You have been given the requirement that the data collected from the distribution centers must be encrypted at rest. You also have to load your distribution center data into your company’s analytics EMR cluster on a daily basis so that your management team can produce daily Key Performance Indicators (KPIs) for the various regional distribution centers.

Which option best meets your encryption at rest requirement?

Select 1 option(s):

Use an AWS KMS customer master key (CMK) for server side encryption when writing your distribution center performance data to S3. Create an IAM role that allows your analytics EMR cluster to have permission to access your S3 buckets and to use the AWS KMS CMK.

Use a customer provided key (SSE-C) for server side encryption when writing your distribution center performance data to S3. Create an IAM role that allows your analytics EMR cluster to have permission to access your S3 buckets and to use the SSE-C key.

Use client side encryption before you write your distribution center performance data to your S3 buckets. Save the client side encryption key in your analytics EMR cluster. Create an IAM role that allows your analytics EMR cluster to have permission to access your S3 buckets and to use the client side key.

Write your distribution center performance data to S3 and encrypt the data using a Kinesis Analytics job as it’s being written to the S3 buckets using server side encryption. Save the server side encryption key in one of your S3 buckets. Create an IAM role that allows your analytics EMR cluster to have permission to access your S3 buckets and to use the server side key.

Q40. You work as a data architect for a sports media data provider. Your company supplies sports data to sports gambling and sports gaming companies. These partner companies use the data your company provides to give their applications the detailed sports information needed to create reliable betting and realistic game simulation. These partners distribute their product as web and mobile applications. Your company currently gathers the data needed to create your sports data media content through a set of EC2 instances running in an auto-scaling group in your AWS account. All of the real-time ingestion, transformation, processing, and visualization of the data for your internal analysts is completed on these EC2 instances.

You need to improve this architecture by decoupling the real-time data collection system components because your company frequently experiences failures where important data is lost.
Which is the most cost effective and performant way to improve your architecture while decoupling your data collection components?

Select 1 option(s):

Load your sports data into S3 using Storage Gateway. Use EMR running Spark to ingest the data into your EMR cluster. Use the Redshift INSERT command to load the data from EMR into your Redshift cluster.

Load your sports data into S3 using Snowball@Edge. Use EMR running Spark to ingest the data into your EMR cluster. Use the Redshift UPDATE command to load the data from EMR into your Redshift cluster.

Load your sports data into S3 using Kinesis Data Firehose. Use EMR running Spark to ingest the data into your EMR cluster. Use the Redshift INSERT command to load the data from EMR into your Redshift cluster.

Load your sports data into S3 using Kinesis Data Firehose. Use EMR running Spark to ingest the data into your EMR cluster. Use the Redshift COPY command to load the data from EMR into your Redshift cluster.

Q41. You work as a data scientist for a government agency contracting firm that collects real-time polling data for various elections and public opinion items. You have built a streaming data collection architecture using Kinesis Data Streams and its Kinesis Producer Library (KPL). Your producer code is using the addUserRecord API call to add records which are eventually flushed to your Kinesis Data stream using the PutRecords API call. You have used the default settings for your PutRecords API KPL calls. Your Kinesis Data Stream PutRecords API call is occasionally experiencing partial and sometimes full failures. You have noticed that your data collection system sometimes experiences excessive retries, sometimes referred to as “retry spamming.”

What is the best approach to mitigate the request spamming resulting from your PutRecords retries?

Select 1 option(s):

Implement rate limiting for the offending producer by setting the token limit to 50% higher than the shard limit.

Expand the capacity of your Kinesis Data Stream while also implementing a suitable partition key strategy

Implement rate limiting for the offending producer by setting the threshold to 50% higher than the shard limit.

Implement rate limiting for the offending producer by setting the threshold to 30% higher than the shard limit.

Q42. You work as a data scientist for a data analytics firm that collects data for various industries, including the airline industry. Your airline clients wish to have your firm create analytics for use in machine learning models that predict air travel in the global market. To this end, you have created a Kinesis Data Streams data collection system that gathers flight data for use in your analysis.

You are writing a consumer application, using the Kinesis Client Library (KCL), that will consume the flight data stream records and process them before placing the data into your S3 data lake.
You need to handle the condition of when your consumer application fails in the middle of reading a data record from the data stream. What is the most efficient way to handle this condition?

Select 1 option(s):

Use a Lambda function to monitor the KCL consumer application’s log and restart the consumer application on failure

Use the KCL application state tracking feature implemented in the global DynamoDB table

Use the KCL application state tracking feature implemented in the DynamoDB table associated with the KCL application that failed when reading

Use the KCL application state tracking feature implemented in the DynamoDB table associated with the shard that the KCL consumer application was reading from when it failed

Q43. You work as a data scientist for a financial services firm that is building an automated trading system using data streamed from market data sources. The market data records the market data sources produce are small in size (512 bytes) and are sent very rapidly (1,500 records per second) to your Kinesis Producer Library based producer application. You have your data collection system configured like this:

Market data source -> KPL producer application -> Kinesis Data stream -> Kinesis Data Firehose stream -> Lambda -> S3
Your Lambda function transforms the market data for use in your automated trading system.
At the size and rate of production of your market data records, your data collection pipeline is constrained. Why is it constrained, and what can you do to remove the constraint?

Select 1 option(s):

The pipeline is constrained because a Kinesis Data Streams shard has a limit of supporting up to 1,000 records per second, or 1 MB throughput. Increase the throughput by using KPL aggregation to aggregate the 1,500 market data records into 10 Kinesis Data Stream records, bringing the records per second to 10, each of which will hold 76 KB.

The pipeline is constrained because you can only have 50 Kinesis Data Firehose delivery streams per region. Increase the throughput by using Service Quota to request a quota increase.

The pipeline is constrained because your Lambda function has a limit of only 15 minutes runtime. Increase the Lambda function capacity by leveraging the messaging fanout pattern for Lambda using SNS.

The pipeline is constrained because you are attempting to send greater than 1 MB of data per second through your shard. Compress your records in your KPL code before sending your records to your Kinesis Data Streams stream. This will remove the message volume constraint.

Q44. You work as a data scientist for an oil refining company. Your team is building a data lake in S3 which will be used to do complex analysis of crude oil chemical compounds. Your company will use this analysis to improve their product to make it more cost effective. Your data lake will have many disparate sources of compound data that need to be loaded into your S3 buckets. You have decided to use AWS Glue to crawl your data sources to allow for the use of Glue transform jobs to process the data while loading it into your S3 buckets.

When you run your Glue crawler on one of your RDS instances you are getting a resource unavailable error. What might be the root cause of this problem?

Select 1 option(s):

The ApplyMapping transform is able to map the source columns from your DynamicFrame to your target columns.

You have tried to use a JDBC connection to access your RDS instance. Glue crawlers connect to RDS instances using the RDS native interface.

You have not opened up all ports to TCP on your database security group.

Your custom classifier used by your Glue crawler is not recognizing the structure of your RDS data.

Q45. You work as a data scientist for an alternative energy source company. Your company has several fields of wind turbines located across several continents and in the open Atlantic and Pacific oceans. You are responsible for implementing the data collection system that feeds turbine sensor data to your analytics platform in real-time for use in preventative maintenance analytics applications for the turbines. These analytics applications are used to schedule maintenance in response to changes in the turbine sensor data in real-time. This allows your company to address turbine low output situations, thereby helping to maximize revenue.

Which data collection architecture handles the frequency, volume, and source of your data while also delivering the real-time analytics needed by the turbine preventive maintenance analytics application in the most cost effective manner?

Select 1 option(s):

Sensor data -> AWS IoT -> S3 -> Lambda -> AWS Batch -> EMR -> Turbine Analytics App

Sensor data -> AWS IoT -> S3 -> Lambda -> Kinesis Data Firehose -> Redshift -> Turbine Analytics App

Sensor data -> AWS IoT -> Kinesis Data Firehose -> Lambda -> Redshift -> Turbine Analytics App

Sensor data -> AWS IoT -> Kinesis Data Firehose -> Lambda -> DynamoDB -> Turbine Analytics App

Q46. You work as a data scientist for a utility company that is implementing real-time management of its electricity meters at the homes of its customers. These meters have sensors on them that transmit usage and other measurements back to your data collection system using AWS IoT. Your management team wishes to use this IoT data to perform analytics and build Key Performance Indicator (KPI) dashboards to help give better service to their customers.

You do not need to transform the IoT data before feeding it into your Redshift cluster. Which architecture option is the most cost effective and efficient?

Select 1 option(s):

Sensor data -> AWS IoT -> KPL Application -> Redshift -> BI Tools

Sensor data -> AWS IoT -> S3 -> Lambda -> Kinesis Data Firehose -> Redshift -> BI Tools

Sensor data -> AWS IoT -> Kinesis Data Firehose -> Lambda -> Redshift -> BI Tools

Sensor data -> AWS IoT -> Kinesis Data Firehose -> S3 -> Redshift -> BI Tools

Q47. You work as a lead data scientist on a development team in a large consulting firm. Your team is working on a contract for a client that needs to gather key statistics from their application server logs. This data needs to be loaded into their S3 data lake for use in analytics applications.

Your data collection process requires transformation of the streamed data records as they are ingested through the collection process. You also have the requirement to keep an unaltered copy of every source record ingested by your data collection process.
Which option meets all of your requirements in the most efficient manner?

Select 1 option(s):

Kinesis Data Stream KPL application streams source data into Kinesis Data Firehose. Kinesis Data Firehose invokes a Lambda function which transforms the data record. Kinesis Data Firehose then writes the transformed record to the S3 data lake destination. Kinesis Data Firehose saves the unaltered record to another S3 destination.

Amazon Kinesis Agent streams source data into Kinesis Data Firehose. Kinesis Data Firehose invokes a Lambda function which transforms the data record. Kinesis Data Firehose then writes the transformed record to the S3 data lake destination. Kinesis Data Firehose saves the unaltered record to another S3 destination.

Amazon Kinesis Agent streams source data into Kinesis Data Firehose. Kinesis Data Firehose invokes a Lambda function which transforms the data record. The Lambda function then writes the transformed record to the S3 data lake destination. Kinesis Data Firehose saves the unaltered record to another S3 destination.

Amazon Kinesis Agent streams source data into Kinesis Data Firehose. Kinesis Data Firehose transforms the data record and writes the transformed record to the S3 data lake destination. Kinesis Data Firehose saves the unaltered record to another S3 destination.

Q48. You work as a data scientist on a development team in a large bank. The bank has asked your team to move your existing on-prem customer account Oracle database to a PostgreSQL RDS instance running in your company’s AWS account. In the process of moving your on-prem database to your AWS RDS instance you cannot take your Oracle database offline. Also, you need to perform some transformations of your database schema as you move to your new instance, including adding new primary keys to certain columns of the customer table and changing some of the data types of some of the target columns.

How would you use AWS services to accomplish this transformation in the most efficient manner?

Select 1 option(s):

Create a Glue crawler that crawls the on-prem Oracle instance and catalogs the schema of the customers data. Then write a Glue ETL job that extracts the data from the Oracle instance, transforms the tables and creates the primary key, and finally loads the data into your PostgreSQL RDS instance.

Create an AWS Schema Conversion Tool (SCT) project that defines the transformations you need to accomplish. Use the AWS SCT to copy your on-prem Oracle database to your RDS PostgreSQL database.

Use the AWS Database Migration Service (DMS) to migrate your Oracle database to your PostgreSQL RDS instance. Use the ongoing replication change data capture (CDC) feature of DMS to allow the migration to capture ongoing changes to your source database as you copy to your target database. Perform your transformations using the table-mapping rules defined in your DMS tasks.

Use an Amazon Kinesis Agent to stream your source data into Kinesis Data Firehose. Kinesis Data Firehose transforms the data record and writes the transformed data into your RDS PostgreSQL instance.

Q49. You work as a data scientist for a farming collective that gathers data from small farms all across the country. The data is used in analytics applications to help the farmer members of the collective better understand their market. Your team has been assigned the task of migrating the collective’s legacy database to an RDS Aurora database on AWS. Since your collective is expanding into other parts of the globe, you want to take advantage of Aurora’s global database feature.

You are running a full load and ongoing replication change data capture (CDC) task to migrate the data from your legacy database instance to your new RDS instance. However, the migration tasks are running very slowly. What is a probable reason for the slowness and what can you do to correct it? (SELECT TWO)

Select 2 option(s):

The process of capturing changes from the source legacy database instance is suboptimal. Use CloudWatch metrics to monitor CDCLatencySource to find source activity that is interfering with your DMS performance, such as if a transaction has started, but has not committed.

The source legacy database must be of the same engine type as the target database engine type. Change your target database to the same instance type as your legacy database instance type.

Your VPC database subnet group is only configured to use two availability zones. Add an additional availability zone to the subnet group.

Your DMS replication instance is running on an EC2 instance in your default VPC.

The process of applying change events to the target instance is suboptimal. Use CloudWatch metrics to monitor CDCLatencyTarget to find target issues that are interfering with your DMS performance, such as lack of primary keys or indexes on the target.

Q50. You work as a data scientist for a streaming music service. Your company wishes to catalog and analyze the metadata about the most frequently streamed songs in their catalog. To do this you have created a Glue crawler that you have scheduled to crawl the company song database every hour. You want to load the song play statistics and metadata into your Redshift data warehouse using a Glue ETL job as soon as the crawler completes.

What is the most efficient way to automatically start the Glue ETL job as soon as the crawler completes?

Select 1 option(s):

Create a Glue Workflow. Create two triggers, one for the crawler and one for the ETL job. The workflow will start the crawler automatically based on a timer trigger. Once the crawler completes, the second trigger, an event trigger, will watch for the event of the crawler completing and start the ETL job. The ETL job will transform and load the data into your Redshift cluster.

Create a Glue trigger, a Lambda function, and a CloudWatch events rule. The Glue trigger is a timed based event that triggers every hour and starts the crawler. The CloudWatch event watches for the crawler state change. When the crawler state changes to complete, the CloudWatch event automatically triggers the Lambda function which starts the Glue ETL job. The ETL job will transform and load the data into your Redshift cluster.

Create two Glue triggers. The first Glue trigger is a timed based event that triggers every hour and starts the crawler. The second Glue trigger watches for the crawler to reach the COMPLETED state and then starts the ETL job. The ETL job will transform and load the data into your Redshift cluster.

Create two Glue triggers. The first Glue trigger is a timer based event that triggers every hour and starts the crawler. The second Glue trigger watches for the crawler to reach the SUCCEEDED state and then starts the ETL job. The ETL job will transform and load the data into your Redshift cluster.

AWS Certified Data Analytics Test 1

Q3. You are working on a project where you need to perform real-time analytics on your application server logs. Your application is split across several EC2 servers in an auto-scaling group and is behind an application load balancer as depicted in this diagram:

Q5. You are a data scientist working for a rental car company that has fleets of rental cars across the globe. Each car is equipped with IoT sensors that report important information about the car’s functioning, location, service levels, mileage, etc.

Q8. You are a data scientist working for a large city that has implemented an electric scooter ride sharing system. Each electric scooter is equipped with IoT sensors that report the scooter’s location, whether it is currently rented out, current renter, battery level, speed of travel, etc.

Q26. You work as a data scientist for a medical data processing company. Your company receives patient data via file feeds into one of your S3 buckets. The data is formatted as a nested JSON document similar to this:

Q47. You work as a lead data scientist on a development team in a large consulting firm. Your team is working on a contract for a client that needs to gather key statistics from their application server logs. This data needs to be loaded into their S3 data lake for use in analytics applications.

Company

Some Useful Links

Our Services

Oh yeah, we're on social media too!