Practical Predictive Analytics

Parallel computing

Parallel computing refers to the ability to perform separate computing tasks at the same time. It does this by distributing the work load. For example if a dataframe needs to simply be sorted by last name, the cluster manager can pass the various names starting with 26 letters of the alphabet to 26 different nodes, have them each sort their own letter, and then ask for them all to be returned to one place where 1 final sort will be performed. This can be quicker than if only 1 machine performed the sort.