Workload and computational requirements

While the previous two areas cover the sizing of the cluster, the workload requirements drive the computational capabilities of the cluster. All CPU-intensive operations require a higher count of CPUs and better configuration for computing. The number of Mapper and Reducer jobs that are run as a part of Hadoop also contribute to the requirements. Mapper tasks are usually higher than Reducer tasks, for example. The ratio of Mapper and Reducer is determined by processing requirements at both ends. 

There is no definitive count that one can reach regarding memory and CPU requirements, as they vary based on replicas of block, the computational processing of tasks, and data storage needs. To help with this, we have provided a calculator which considers different configurations of a Hadoop cluster, such as CPU-intensive, memory-intensive, and balanced.