AWS Certified Solutions Architect – Professional — Question 937
A company uses an Amazon EMR cluster to process data once a day. The raw data comes from Amazon S3, and the resulting processed data is also stored in
Amazon S3. The processing must complete within 4 hours; currently, it only takes 3 hours. However, the processing time is taking 5 to 10 minutes longer each week due to an increasing volume of raw data.
The team is also concerned about rising costs as the compute capacity increases. The EMR cluster is currently running on three m3.xlarge instances (one master and two core nodes).
Which of the following solutions will reduce costs related to the increasing compute needs?
Answer options
- A. Add additional task nodes, but have the team purchase an all-upfront convertible Reserved Instance for each additional node to offset the costs.
- B. Add additional task nodes, but use instance fleets with the master node in on-Demand mode and a mix of On-Demand and Spot Instances for the core and task nodes. Purchase a scheduled Reserved Instance for the master node.
- C. Add additional task nodes, but use instance fleets with the master node in Spot mode and a mix of On-Demand and Spot Instances for the core and task nodes. Purchase enough scheduled Reserved Instances to offset the cost of running any On-Demand instances.
- D. Add additional task nodes, but use instance fleets with the master node in On-Demand mode and a mix of On-Demand and Spot Instances for the core and task nodes. Purchase a standard all-upfront Reserved Instance for the master node.
Correct answer: B
Explanation
Using a scheduled Reserved Instance for the master node is highly cost-effective because the EMR cluster only runs for a few hours daily, making a 24/7 standard RI wasteful. Spot instances should be leveraged for core and task nodes to minimize compute costs, while the master node must remain On-Demand to prevent cluster termination from Spot interruptions. Therefore, combining instance fleets with Spot instances and a scheduled RI for the master node provides the optimal balance of cost savings and cluster stability.