AWS Certified Data Engineer Associate · Free Practice Question Medium
Question 4
A business analytics team is tasked with configuring a provisioned Amazon EMR cluster optimized for executing Apache Spark jobs to analyze extensive datasets. The team must ensure that the cluster operates cost-effectively without compromising on performance or reliability.
To achieve this, what two recommendations should the team follow to configure their Amazon EMR resources? (Select TWO.)
-
A
Use Graviton instances for core and task nodes to leverage better price-performance.
-
B
Choose Spot Instances for task nodes to take advantage of lower prices for flexible workloads.
-
C
Configure Amazon EMR to use Amazon S3 as a data lake for durable and cost-effective storage.
-
D
Implement EMR Managed Scaling to automatically resize the cluster based on workload.
-
E
Select Reserved Instances for core nodes to reduce costs with long-term commitment.
Reveal correct answers
Correct answers: A, C
Explanation
Using Amazon S3 as a data lake for Amazon EMR is a common best practice because S3 provides highly durable storage at a lower cost compared to HDFS on EMR. S3 also decouples storage from compute, allowing teams to shut down EMR clusters when not in use and avoid paying for persistent HDFS on EMR nodes, thus optimizing costs.
AWS Graviton instances, which are powered by Arm-based processors, offer a better price-performance ratio compared to traditional x86-based instances. They are designed to deliver cost savings and are optimized for performance, making them a good choice for running cost-optimized and performance-intensive big data workloads on EMR.
CORRECT: "Configure Amazon EMR to use Amazon S3 as a data lake for durable and cost-effective storage" is a correct answer (as explained above.)
CORRECT: "Use Graviton instances for core and task nodes to leverage better price-performance" is also a correct answer (as explained above.)
INCORRECT: "Implement EMR Managed Scaling to automatically resize the cluster based on workload" is incorrect.
While HDFS is the traditional storage system used by Hadoop and EMR, it requires the cluster to be continuously running, which can incur higher costs compared to using Amazon S3 for storage, where compute and storage can be scaled independently.
INCORRECT: "Select Reserved Instances for core nodes to reduce costs with long-term commitment" is incorrect.
x86-based instances may not be the most cost-effective option compared to AWS Graviton instances when price-performance is considered, especially for the long-running and resource-intensive workloads typical in big data analysis.
INCORRECT: "Choose Spot Instances for task nodes to take advantage of lower prices for flexible workloads" is incorrect.
Spot Instances can offer significant savings for fault-tolerant and flexible workloads. However, using Spot Instances for all primary nodes (which include Master nodes) is not recommended due to the possibility of Spot Instance termination by AWS if there is higher demand for the capacity, which can affect cluster reliability and continuity of long-running jobs. It's more common to use Spot Instances for task nodes and not for core or master nodes which are critical for the cluster's operation.
References:
https://docs.aws.amazon.com/emr/latest/ManagementGuide/emr-plan-storage.html
https://aws.amazon.com/ec2/instance-types/
Save time with our AWS cheat sheets:
https://digitalcloud.training/aws-certified-data-engineer-associate-cheat-sheet/
Discussion
Think the marked answer is wrong, or have a better explanation? Share it below — comments appear after review.
