The interaction between hadoop hdfs block size and linux file system block size

2
votes

I understood hadoop block size is 64MB and linux FS is 4KB. My understanding from reading is that hadoop hdfs work on top of linux FS itself.

How does hadoop file system actually work with linux 4KB block size? Does 64MB block get broken down to 4KB blocks and saved to the disk during write operation, for example?

1
hadoophdfs

1 Answers

5
votes

Yes, what you said is correct. HDFS is just an abstraction above Linux FS where a block of 64MB is stored in continuous blocks of 4KB linux FS to have fast access. When you write 64MB of HDFS block will be treated as a single file by Linux and internally it breaks it down to 4KB blocks.