0

Hadoop file system is physical file system or virtual file system

user2183044
  • 33
  • 2
  • 6

1 Answers1

1

HDFS is a virtual FS that runs on top of your physical FS.

Tariq
  • 34,076
  • 8
  • 57
  • 79
  • why HDFS default block size is 64MB? why not 32MB or 50MB? i knew its 1024 rule based? but then it can be 32MB also right? Please justify.. – user2183044 May 06 '13 at 03:26
  • if the block size is too small or too big it'll become an overhead on your network and disk. if the blocks are very small, it will increase the disk i/o and the network traffic. and if it is very huge it'll again cause network congestion. also you might face memory related issues in long running tasks. so the block size is kept "moderately" high(normally 64M or 128M) – Tariq May 06 '13 at 11:19
  • Can anyone define what is HDFS Client? – user2183044 May 15 '13 at 16:11
  • I have few More Questions.1) Assume i am loading 1GB data into HDFS having block default block size(64MB), after 10secs 128MB of data is written into assume 2 data nodes(2 blocks), the write operation is still performing, can i read those written blocks? if so how can i read? what is the command for that. 2) can i have parallel writing/reading files into HDFS? like two write jobs or two read jobs? – user2183044 May 15 '13 at 16:16
  • @Tariq: Correct me if I am wrong.HDFS is a logical file system. but why we are saying in storage perspective Block is a physical representation data and Split is a logical representation of data. At Os level, we can't see any physical file. so how Block is physical data... –  Oct 16 '16 at 04:27