← Big Data & Hadoop learning pathBig Data & Hadoop · Module 3 of 10
HDFS
Store huge files reliably across a cluster.
📖 Key ideas
- Files split into large blocks (default 128 MB).
- Each block is replicated (default 3 copies) on different nodes.
- NameNode stores metadata; DataNodes store blocks.
- Write once, read many access pattern.
- Commands: hdfs dfs -put, -ls, -cat, -get.
💡 Real-life example: HDFS is tearing a big book into chapters and keeping photocopies of each chapter in three libraries.
BlockChunk of a file in HDFS
Replication factorNumber of copies per block
NameNodeMaster holding metadata
DataNodeWorker storing blocks