Amazon EMR Instance Types

Correct Answer: m5.xlarge

For running Spark and Hadoop jobs with heavy workloads, the m5.xlarge instance type is the best choice among the given options. The 'm' series instances are general-purpose instances that provide a balance of compute, memory, and network resources. The 'm5' family, in particular, is built on the AWS Nitro System, offering better performance and higher bandwidth. The 'xlarge' size provides sufficient resources to handle the computational and memory demands of Spark and Hadoop jobs. Other options like t2.micro or t3.large are burstable instances not suitable for consistent heavy workloads, while c4.large is compute-optimized but may lack the memory required for big data processing. The a1.large is ARM-based and not ideal for traditional Hadoop/Spark workloads, and r5.large, while memory-optimized, may not provide the best balance for these specific jobs.

Correct Answer: c5.xlarge and r5.xlarge

c5.xlarge instances are optimized for compute-intensive workloads, while r5.xlarge instances are optimized for memory-intensive workloads. Combining these two instance types will efficiently optimize both computing and memory resources.

Correct Answer: m5.large

For EMR task nodes running mixed Spark workloads on x86_64, m5.large provides the best balance of CPU, memory (2 vCPU, 8 GiB), network, and price among the given options, delivering consistent performance without CPU credits. t2.micro is unsuitable due to limited resources and burstable CPU credits. c4.large is an older compute-optimized instance with less memory per vCPU and generally inferior price/performance for mixed workloads. a1.large uses ARM (first-generation Graviton), which can introduce compatibility and performance trade-offs for EMR stacks built for x86_64. Therefore, m5.large is the most cost-effective choice without sacrificing performance.