MapReduce (MR) is a programming paradigm at the core of Hadoop. It can scale the processing of data to massively high volumes. The data and processing can be distributed to hundreds and thousands of nodes for horizontal scalability. As the name suggests, the MR jobs contain two phases:
- The map phase
- The reduce phase
In the map phase, the dataset is divided into chunks and sent to an independent process to gather the result. These parallel mapper processes work independently on various available nodes in the cluster. Once their processing is completed (map task), the results are shuffled and sorted before initiating the reduce tasks. The reduce tasks once again run independently on the available nodes and the entire computation ...