If each block is saved multiple times for fault tolerance (say 3 times), then does that mean that a 100TB datafile actually takes 300 TB size on the HDFS?
If each block is saved multiple times for fault tolerance (say 3 times), then does that mean that a 100TB datafile actually takes 300 TB size on the HDFS?
Yes, that is correct. The data must be stored on 3 different data nodes, and therefore it duplicated 3 times.