The original open-source framework for distributed storage and batch processing of very large datasets across commodity machines.
The three pieces
HDFS — distributed file system that splits and replicates blocks across nodes.
YARN — cluster resource manager and job scheduler.
MapReduce — the batch programming model: map to transform, reduce to aggregate.
On AWS
Amazon EMR is managed Hadoop — AWS provisions and operates the cluster; EMR clusters usually read from Amazon S3 (via EMRFS) rather than HDFS, which decouples storage from compute.