SOA-C02 Cost and Performance Optimization Practice Question
A company runs a batch processing job on Amazon EMR every night. The job runs for 6 hours and requires a cluster of 20 m5.xlarge instances. The company wants to reduce costs while ensuring the job completes on time. Which solution is MOST cost-effective?
⚠ Common exam trap
SOA-C02 often tests the misconception that Spot Instances can be used for all node types, but the primary node must be On-Demand to avoid cluster termination, and core nodes may risk data loss if not using EMRFS.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Use Spot Instances for core and task nodes and an On-Demand instance for the primary node.
Amazon EMR clusters consist of a primary node (master), core nodes (which run HDFS and task processes), and task nodes (which only run tasks and can be lost without data loss). Spot Instances are ideal for task nodes because they can be interrupted without affecting HDFS data, and for core nodes if the cluster is resilient to interruptions (e.g., using EMRFS consistent view or if the job can tolerate some core node loss). The primary node must be On-Demand to avoid cluster termination if the Spot Instance is reclaimed. This mix minimizes cost while ensuring the job completes on time.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Use On-Demand instances for all nodes.
Why it's wrong here
Running every node as On-Demand fails to leverage Amazon EMR's most significant cost lever for transient workloads. Since the job runs nightly for only about 6 hours, the cluster is idle 75% of the day, making On-Demand's premium pricing a poor fit. On-Demand provides no discount for short-lived, interruption-tolerant compute, so this choice maximizes cost without any reliability benefit over a mixed strategy.
- ✗
Purchase Reserved Instances for the entire cluster.
Why it's wrong here
Reserved Instances require a 1- or 3-year commitment and are billed hourly regardless of usage, which aligns poorly with a 6-hour nightly job that only runs ~2,190 hours per year—far below the 8,760 hours that make RIs cost-effective. You would pay for the full reservation term even though the cluster is stopped for most of the day. Furthermore, EMR's transient cluster pattern typically renders RI utilization low, and any change in job frequency would strand the unused capacity.
- ✓
Use Spot Instances for core and task nodes and an On-Demand instance for the primary node.
Why this is correct
This is the AWS-recommended cost-optimization pattern for transient EMR clusters: the primary node runs On-Demand to guarantee stable HDFS NameNode and ResourceManager availability, while core and task nodes run Spot Instances to exploit steep discounts (often 50–90% off On-Demand). Spot interruptions on core and task nodes are recoverable by EMR's instance-group resizing and task-node retries, but losing the primary node would fail the entire cluster. This mix preserves durability and responsiveness for the critical coordinator while dramatically lowering compute cost for the bulk of the cluster.
- ✗
Use Spot Instances for the primary node and On-Demand for core and task nodes.
Why it's wrong here
Putting the primary node on Spot is a critical anti-pattern because the primary node hosts the NameNode (or HA services) and the ResourceManager; if Spot reclaims it, the cluster immediately fails and all work is lost, including any partially written HDFS data. Core and task nodes, on the other hand, can often tolerate Spot interruptions through EMR's automatic replacements and the job's ability to retry failed tasks. Therefore, this option inverts the correct risk profile—it exposes the most sensitive component to the least reliable instance type while paying a premium for nodes that are naturally resilient.
Go deeper
Related to this question
About these practice questions
This SOA-C02 question is part of Courseiva's 1,169-question bank — original exam-style content with full explanations and wrong-answer analysis, never real exam questions or exam dumps. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official Amazon Web Services exam blueprint
This SOA-C02 practice question is part of Courseiva's free Amazon Web Services certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the SOA-C02 exam.