In this recipe, we explore some of the important configuration options of Hadoop YARN and Hadoop MapReduce. Commercial Hadoop distributions typically provide a GUI-based approach to specify Hadoop configurations.
YARN allocates resource containers to the applications based on the resource requests made by the applications and the available resource capacity of the cluster. A resource request by an application would consist of the number of containers required and the resource requirement of each container. Currently, most container resource requirements are specified using the amount of memory. Hence, our focus in this recipe will be mainly on configuring the memory allocation of a YARN cluster.