Hadoop File System Architecture Single Namespace for entire NameNode Metadata(name,replicas,..) cluster Metadata Ops Files are broken up into BackupNode blocks Metadata(name,replicas..) Client Typically 64 MB block size Block Read DataNodes Each block replicated on multiple DataNodes L Blocks Client 。 Finds location of blocks Client from NameNode Block Write Accesses data directly Replication from DataNode Rack 1 Rack 2 Database System Concepts-7th Edition 10.6 @Silberschatz,Korth and SudarshanDatabase System Concepts - 7 10.6 ©Silberschatz, Korth and Sudarshan th Edition Hadoop File System Architecture ▪ Single Namespace for entire cluster ▪ Files are broken up into blocks • Typically 64 MB block size • Each block replicated on multiple DataNodes ▪ Client • Finds location of blocks from NameNode • Accesses data directly from DataNode