I am setting up a new virtualized server with KVM and LVM. The KVM host and all guests will be installed on a SSD drive. Each guest will have access to a RAID10 array made of 4x3TB sata drives, used for data and storage.
My problem is that I have to much space (6TB useable), which I don't really need at the moment.
One approach, first, is to make one big 6TB space with mdadm and then create a big LVM volume group which can then be "splitted" by creating logical volumes.
Another approach, second, would be to first split up each hard drive (3TB) into, lets say, four 0.75TB primary partitions (sdx1,sdx2,sdx3,sdx4 , x=a,b,c,d) with fdisk. Then I could use
mdadm --verbose --create /dev/md1 --level=10 --chunk=256 --raid-devices=4 /dev/sda1 /dev/sdb1 /dev/sdc1 /dev/sdd1
In this case, the /dev/md1 would be 1.5TB RAID10 array, which can be formed into a LVM volume group followed by the creation of logical volumes.
If I in the future need more space I can just create /dev/md2 and so on. The /dev/md1, /dev/md2, /dev/md3 and /dev/md4 equals the 6TB.
I guess the first method can be called a top-down approach (going from big to small). The second method is perhaps a buttom-up approach (going from small to big).
This is the first time using RAID and I wonder if there are any drawbacks using the second approach? Most people seems to be using the first approach, or? So what is the best way to set up my RAID10? Perhaps there is no difference, just different ways to look at the same thing? How would you do it?