At the organization I work for we have a hadoop cluster. We recently purchased new datanodes to expand the cluster. The new datanodes have 12 disks (8TB) instead of 10 (6TB), as the older datanodes.
After some research (mainly this community) I came to understand that its not a requirement to have the same number of drives in each node or equally sized disks.
On each old datanode we have 8 hdfs partitions one for every disk.
Should we create 10 hdfs partitions on each new datanode or create 8 volumes using the 10 disks?
Should we utilize all the available storage?
And if we do is there something specific we should monitor?
On each old datanode we have 10 hdfs partitions one for every disk.
Should we create 12 hdfs partitions on each new datanode or create 10 volumes using the 12 disks?