AWS Certified Data Engineer Associate DEA-C01 Practice Question

Using EC2 instance store volumes provides high-throughput, low-latency storage that is included in the hourly price of the instance, so no separate EBS charges are incurred. Because the data is temporary, the loss of instance store when the cluster is terminated is not an issue. Creating a transient cluster each night ensures that storage exists only for the duration of the job, eliminating the recurring EBS cost while maintaining similar performance.

Storing data exclusively in Amazon S3 through EMRFS removes EBS cost but adds network latency and S3 request overhead, which can lengthen Spark shuffle phases. Re-platforming to AWS Glue might reduce management overhead but can be more expensive for very large, shuffle-heavy workloads and would require significant refactoring. Switching to low-cost sc1 EBS volumes lowers price but also drastically reduces throughput, likely increasing job runtimes.