A manufacturing company has been collecting IoT sensor data from devices on its factory floor for a year and is storing the data in Amazon Redshift for daily analysis. A data analyst has determined that, at an expected ingestion rate of about 2 TB per day, the cluster will be undersized in less than 4 months. A long-term solution is needed. The data analyst has indicated that most queries only reference the most recent 13 months of data, yet there are also quarterly reports that need to query all the data generated from the past 7 years. The chief technology officer (CTO) is concerned about the costs, administrative effort, and performance of a long-term solution.
Which solution should the data analyst use to meet these requirements?
- A. Create a daily job in AWS Glue to UNLOAD records older than 13 months to Amazon S3 and delete those records from Amazon Redshift. Create an external table in Amazon Redshift to point to the S3 location. Use Amazon Redshift Spectrum to join to data that is older than 13 months.
- B. Take a snapshot of the Amazon Redshift cluster. Restore the cluster to a new cluster using dense storage nodes with additional storage capacity.
- C. Execute a CREATE TABLE AS SELECT (CTAS) statement to move records that are older than 13 months to quarterly partitioned data in Amazon Redshift Spectrum backed by Amazon S3.
- D. Unload all the tables in Amazon Redshift to an Amazon S3 bucket using S3 Intelligent-Tiering. Use AWS Glue to crawl the S3 bucket location to create external tables in an AWS Glue Data Catalog. Create an Amazon EMR cluster using Auto Scaling for any daily analytics needs, and use Amazon Athena for the quarterly reports, with both using the same AWS Glue Data Catalog.
B = wrong, this will not solve either cost or scale problem. C = wrong, to create table on S3 you use CREATE EXTERNAL TABLE not CTAS, also this does not remove older data. D = wrong, nonsense.
An insurance company has raw data in JSON format that is sent without a predefined schedule through an Amazon Kinesis Data Firehose delivery stream to an
Amazon S3 bucket. An AWS Glue crawler is scheduled to run every 8 hours to update the schema in the data catalog of the tables stored in the S3 bucket. Data analysts analyze the data using Apache Spark SQL on Amazon EMR set up with AWS Glue Data Catalog as the metastore. Data analysts say that, occasionally, the data they receive is stale. A data engineer needs to provide access to the most up-to-date data.
Which solution meets these requirements?
- A. Create an external schema based on the AWS Glue Data Catalog on the existing Amazon Redshift cluster to query new data in Amazon S3 with Amazon Redshift Spectrum.
- B. Use Amazon CloudWatch Events with the rate (1 hour) expression to execute the AWS Glue crawler every hour.
- C. Using the AWS CLI, modify the execution schedule of the AWS Glue crawler from 8 hours to 1 minute.
- D. Run the AWS Glue crawler from an AWS Lambda function triggered by an S3:ObjectCreated:* event notification on the S3 bucket.
Note: This is a textbook question. A = wrong, won’t work because schema will update every 8 hours. B = wrong, not most up-to-date. C = wrong, minimum schedule is 5 minutes.
A company that produces network devices has millions of users. Data is collected from the devices on an hourly basis and stored in an Amazon S3 data lake.
The company runs analyses on the last 24 hours of data flow logs for abnormality detection and to troubleshoot and resolve user issues. The company also analyzes historical logs dating back 2 years to discover patterns and look for improvement opportunities.
The data flow logs contain many metrics, such as date, timestamp, source IP, and target IP. There are about 10 billion events every day.
How should this data be stored for optimal performance?
- A. In Apache ORC partitioned by date and sorted by source IP
- B. In compressed .csv partitioned by date and sorted by source IP
- C. In Apache Parquet partitioned by source IP and sorted by date
- D. In compressed nested JSON partitioned by source IP and sorted by date
B and D = wrong, use columnar format. C = wrong, partition by date so historical data and be separated.
A banking company is currently using an Amazon Redshift cluster with dense storage (DS) nodes to store sensitive data. An audit found that the cluster is unencrypted. Compliance requirements state that a database with sensitive data must be encrypted through a hardware security module (HSM) with automated key rotation.
Which combination of steps is required to achieve compliance? (Choose two.)
- A. Set up a trusted connection with HSM using a client and server certificate with automatic key rotation.
- B. Modify the cluster with an HSM encryption option and automatic key rotation.
- C. Create a new HSM-encrypted Amazon Redshift cluster and migrate the data to the new cluster.
- D. Enable HSM with key rotation through the AWS CLI.
- E. Enable Elliptic Curve Diffie-Hellman Ephemeral (ECDHE) encryption in the HSM.
B = wrong, to use HSM you have to create new clusters. D = wrong, key rotation is not done by HSM, but Redshift. E = wrong, nonsense. This is a textbook question.
A company is planning to do a proof of concept for a machine earning (ML) project using Amazon SageMaker with a subset of existing on-premises data hosted in the company’s 3 TB data warehouse. For part of the project, AWS Direct Connect is established and tested. To prepare the data for ML, data analysts are performing data curation. The data analysts want to perform multiple step, including mapping, dropping null fields, resolving choice, and splitting fields. The company needs the fastest solution to curate the data for this project.
Which solution meets these requirements?
- A. Ingest data into Amazon S3 using AWS DataSync and use Apache Spark scrips to curate the data in an Amazon EMR cluster. Store the curated data in Amazon S3 for ML processing.
- B. Create custom ETL jobs on-premises to curate the data. Use AWS DMS to ingest data into Amazon S3 for ML processing.
- C. Ingest data into Amazon S3 using AWS DMS. Use AWS Glue to perform data curation and store the data in Amazon S3 for ML processing.
- D. Take a full backup of the data store and ship the backup files using AWS Snowball. Upload Snowball data into Amazon S3 and schedule data curation jobs using AWS Batch to prepare the data for ML.
A = wrong, DataSync is for storage migration not data warehouse. B = wrong, ETL job on-premise is not fast. D = wrong, too slow.
A US-based sneaker retail company launched its global website. All the transaction data is stored in Amazon RDS and curated historic transaction data is stored in Amazon Redshift in the us-east-1 Region. The business intelligence (BI) team wants to enhance the user experience by providing a dashboard for sneaker trends.
The BI team decides to use Amazon QuickSight to render the website dashboards. During development, a team in Japan provisioned Amazon QuickSight in ap-northeast-1. The team is having difficulty connecting Amazon QuickSight from ap-northeast-1 to Amazon Redshift in us-east-1.
Which solution will solve this issue and meet the requirements?
- A. In the Amazon Redshift console, choose to configure cross-Region snapshots and set the destination Region as ap-northeast-1. Restore the Amazon Redshift Cluster from the snapshot and connect to Amazon QuickSight launched in ap-northeast-1.
- B. Create a VPC endpoint from the Amazon QuickSight VPC to the Amazon Redshift VPC so Amazon QuickSight can access data from Amazon Redshift.
- C. Create an Amazon Redshift endpoint connection string with Region information in the string and use this connection string in Amazon QuickSight to connect to Amazon Redshift.
- D. Create a new security group for Amazon Redshift in us-east-1 with an inbound rule authorizing access from the appropriate IP address range for the Amazon QuickSight servers in ap-northeast-1.
Note: This is a textbook question. Although it looks like a weird answer, but actually you cannot have cross-region access if you put QuickSight in a VPC.
An airline has .csv-formatted data stored in Amazon S3 with an AWS Glue Data Catalog. Data analysts want to join this data with call center data stored in Amazon Redshift as part of a daily batch process. The Amazon Redshift cluster is already under a heavy load. The solution must be managed, serverless, well-functioning, and minimize the load on the existing Amazon Redshift cluster. The solution should also require minimal effort and development activity.
Which solution meets these requirements?
- A. Unload the call center data from Amazon Redshift to Amazon S3 using an AWS Lambda function. Perform the join with AWS Glue ETL scripts.
- B. Export the call center data from Amazon Redshift using a Python shell in AWS Glue. Perform the join with AWS Glue ETL scripts.
- C. Create an external table using Amazon Redshift Spectrum for the call center data and perform the join with Amazon Redshift.
- D. Export the call center data from Amazon Redshift to Amazon EMR using Apache Sqoop. Perform the join with Apache Hive.
Note: This is a free score question. Remember Redshift Spectrum is serverless.
A data analyst is using Amazon QuickSight for data visualization across multiple datasets generated by applications. Each application stores files within a separate Amazon S3 bucket. AWS Glue Data Catalog is used as a central catalog across all application data in Amazon S3. A new application stores its data within a separate S3 bucket. After updating the catalog to include the new application data source, the data analyst created a new Amazon QuickSight data source from an Amazon Athena table, but the import into SPICE failed.
How should the data analyst resolve the issue?
- A. Edit the permissions for the AWS Glue Data Catalog from within the Amazon QuickSight console.
- B. Edit the permissions for the new S3 bucket from within the Amazon QuickSight console.
- C. Edit the permissions for the AWS Glue Data Catalog from within the AWS Glue console.
- D. Edit the permissions for the new S3 bucket from within the S3 console.
A and C = wrong, Glue is updated without error so no problem there, also Glue permission is not set in QuickSight, QuickSight connects to Athena, not via Glue. D = wrong, QuickSight will create the rules for you.
Note: Athena is a very strange service as it transparently uses user’s access to S3 buckets, instead of relying on service roles like most other AWS services. So, to be able to use Athena, the user itself will need to have S3 access, there is no service role creation for Athena.
A team of data scientists plans to analyze market trend data for their company’s new investment strategy. The trend data comes from five different data sources in large volumes. The team wants to utilize Amazon Kinesis to support their use case. The team uses SQL-like queries to analyze trends and wants to send notifications based on certain significant patterns in the trends. Additionally, the data scientists want to save the data to Amazon S3 for archival and historical re-processing, and use AWS managed services wherever possible. The team wants to implement the lowest-cost solution.
Which solution meets these requirements?
- A. Publish data to one Kinesis data stream. Deploy a custom application using the Kinesis Client Library (KCL) for analyzing trends, and send notifications using Amazon SNS. Configure Kinesis Data Firehose on the Kinesis data stream to persist data to an S3 bucket.
- B. Publish data to one Kinesis data stream. Deploy Kinesis Data Analytic to the stream for analyzing trends, and configure an AWS Lambda function as an output to send notifications using Amazon SNS. Configure Kinesis Data Firehose on the Kinesis data stream to persist data to an S3 bucket.
- C. Publish data to two Kinesis data streams. Deploy Kinesis Data Analytics to the first stream for analyzing trends, and configure an AWS Lambda function as an output to send notifications using Amazon SNS. Configure Kinesis Data Firehose on the second Kinesis data stream to persist data to an S3 bucket.
- D. Publish data to two Kinesis data streams. Deploy a custom application using the Kinesis Client Library (KCL) to the first stream for analyzing trends, and send notifications using Amazon SNS. Configure Kinesis Data Firehose on the second Kinesis data stream to persist data to an S3 bucket.
Note: KCL is self-service, so A and D are out. If you want multiple consumers for the same stream, just make sure you have enough shards to deal with the read-write throughput. A stream is different from a queue in that one can traverse back and forth in a stream, where in a queue one can only process one by one.
A company currently uses Amazon Athena to query its global datasets. The regional data is stored in Amazon S3 in the us-east-1 and us-west-2 Regions. The data is not encrypted. To simplify the query process and manage it centrally, the company wants to use Athena in us-west-2 to query data from Amazon S3 in both
Regions. The solution should be as low-cost as possible.
What should the company do to achieve this goal?
- A. Use AWS DMS to migrate the AWS Glue Data Catalog from us-east-1 to us-west-2. Run Athena queries in us-west-2.
- B. Run the AWS Glue crawler in us-west-2 to catalog datasets in all Regions. Once the data is crawled, run Athena queries in us-west-2.
- C. Enable cross-Region replication for the S3 buckets in us-east-1 to replicate data in us-west-2. Once the data is replicated in us-west-2, run the AWS Glue crawler there to update the AWS Glue Data Catalog in us-west-2 and run Athena queries.
- D. Update AWS Glue resource policies to provide us-east-1 AWS Glue Data Catalog access to us-west-2. Once the catalog in us-west-2 has access to the catalog in us-east-1, run Athena queries in us-west-2.
Note: This one is confusing. Source says Athena supports cross-region query, while official docs says it is not supported. The official docs were updated to reflect its support on cross-region query, so B should be the most cost-effective. C could work but is too costly.
A large company receives files from external parties in Amazon EC2 throughout the day. At the end of the day, the files are combined into a single file, compressed into a gzip file, and uploaded to Amazon S3. The total size of all the files is close to 100 GB daily. Once the files are uploaded to Amazon S3, an
AWS Batch program executes a COPY command to load the files into an Amazon Redshift cluster.
Which program modification will accelerate the COPY process?
- A. Upload the individual files to Amazon S3 and run the COPY command as soon as the files become available.
- B. Split the number of files so they are equal to a multiple of the number of slices in the Amazon Redshift cluster. Gzip and upload the files to Amazon S3. Run the COPY command on the files.
- C. Split the number of files so they are equal to a multiple of the number of compute nodes in the Amazon Redshift cluster. Gzip and upload the files to Amazon S3. Run the COPY command on the files.
- D. Apply sharding by breaking up the files so the distkey columns with the same values go to the same file. Gzip and upload the sharded files to Amazon S3. Run the COPY command on the files.
Note: This is a textbook question. Sequential loading vs. parallel loading.
A large ride-sharing company has thousands of drivers globally serving millions of unique customers every day. The company has decided to migrate an existing data mart to Amazon Redshift. The existing schema includes the following tables.
– A trips fact table for information on completed rides.
– A drivers dimension table for driver profiles.
– A customers fact table holding customer profile information.
The company analyzes trip details by date and destination to examine profitability by region. The drivers data rarely changes. The customers data frequently changes.
What table design provides optimal query performance?
- A. Use DISTSTYLE KEY (destination) for the trips table and sort by date. Use DISTSTYLE ALL for the drivers and customers tables.
- B. Use DISTSTYLE EVEN for the trips table and sort by date. Use DISTSTYLE ALL for the drivers table. Use DISTSTYLE EVEN for the customers table.
- C. Use DISTSTYLE KEY (destination) for the trips table and sort by date. Use DISTSTYLE ALL for the drivers table. Use DISTSTYLE EVEN for the customers table.
- D. Use DISTSTYLE EVEN for the drivers table and sort by date. Use DISTSTYLE ALL for both fact tables.
Note: This is a textbook question. However, if there is an answer where both customer and driver tables are EVEN, I would go for it. ALL does not quite give benefits over EVEN.