Which part of HDFS is recommended to use RAID?

Question

Can anyone tell me which part of HDFS is recommended to use RAID?

Answer 1

No need to use RAID at all. I would suggest configuring Namenode-HA to avoid any single-point-of-failure for HDFS. This will make sure that the Namenode Metadata is written in multiple copies throughout the JournalNodes. Where I see RAID as useful are the partitions for the operating system, logs,.. but this is obviously "below" HDFS.

If you are interested to learn Hadoop, you can check out this Hadoop Certification program by Intellipaat.

Also, watch this video for more information: