How HDFS guarantees data security

1. HDFS is a typical Master/Slave architecture. It is often composed of one NameNode and multiple DataNodes. NamNode is a cluster; and HDFS is stored in blocks. For fault-tolerant files, each block will have a copy.
2. The first copy is generally placed on the node where the client is located (if the client does not have a DataNode, it is placed randomly), the second copy is placed on a different node in the same rack as the first copy, and the third copy is placed on DataNode nodes of different racks, follow the principle of proximity when fetching
3. The DataNode has Block as a unit, reporting the heartbeat status every 3s. If the heartbeat status is not reported within 10 minutes, the NameNode considers the block to be dead, and the NameNode will back up the data on it to another DataNode to ensure the number of copies of the data
4. DataNode will report all block status information on its own node to NameNode every hour by default
5. Safemode mode: DataNode will periodically report Block information and HDFS metadata is backed up by SecondaryNameNode or HA