AWS Certified Solutions Architect – Associate (SAA-C03)Design High-Performing ArchitecturesMedium
A data analytics company needs to process petabytes of historical data stored in Amazon S3. The processing jobs are intermittent, run for several hours, and require significant compute resources. Cost optimization is a major concern. Which compute solution would be most cost-effective and scalable for this workload?
- AEC2 Spot Instances with Amazon EMR
- BOn-Demand EC2 instances in an Auto Scaling group
- CReserved EC2 Instances
- DAWS Fargate for serverless container orchestration
Show answer & explanationAnswer & explanation
Correct answer: A. EC2 Spot Instances with Amazon EMR
Amazon EMR is a managed cluster platform that simplifies running big data frameworks. Combining EMR with EC2 Spot Instances is highly cost-effective for intermittent, fault-tolerant workloads like historical data processing, as Spot Instances can offer significant savings (up to 90%) compared to On-Demand prices.
Why the other options are wrong
- B. On-Demand EC2 instances are billed by the hour/second and are more expensive than Spot Instances, making them less cost-effective for intermittent, non-critical workloads.
- C. Reserved EC2 Instances are suitable for steady-state, predictable workloads that run 24/7, not for intermittent processing jobs, and do not offer the flexibility or cost savings of Spot Instances for this use case.
- D. AWS Fargate is a serverless compute engine for containers, suitable for microservices or web applications, but not typically the most cost-effective or optimized solution for petabyte-scale big data processing jobs when compared to EMR with Spot Instances.
EC2 Spot Instances with Amazon EMR
Leveraging EC2 Spot Instances with Amazon EMR provides a highly cost-effective solution for processing large datasets, ideal for fault-tolerant, intermittent big data workloads.
- Spot Instances offer up to 90% savings.
- EMR simplifies running big data frameworks like Hadoop, Spark.
- Suitable for workloads tolerant to interruptions.
Memory trick: Spot the savings for big data with EMR.