I am quite new to Linux world, please forgive my ignorance...:-). I recently took up a temp job, trying to implement an efficient and error free backup methodology for a 2TB+ Linux file system. The institution is running an IBM X3500 with RHEL 5, connecting to a Dell PowerVault MD3000i iSCSI SAN. The MD3000i has 15 x 1TB Hitachi HUA721010KLA330 1TB Hard Drive. One is used for hot spare. The rest are grouped into two disk groups (RAID-5) : Disk Group 1 and Disk Group 2. Each Disk Group as 7 x 1TB physical drives grouped into 3 virtual disks, with the capacity of 2TB+2TB+1.4TB respectively.
Please refer to this diagram:
Here is how it is organized under RHEL 5:
Disk group 1 (volume group : vg0) is the production database. Disk group 2 (volume group : vg1) is the backup database. The setup was done last year by Dell engineer I think.
My questions are :
- Can I use MD3000i HW snapshot? My impression is that a virtual disk in MD3000i can not be larger than 2TB, therefore the whole production database cannot be contained in a single virtual disk. Does it mean than we cannot use MD3000i HW snapshot?
- Can I use HW snapshot between two disk group? (I think the answer is no?)
- If HW snapshot is not possible, then I need to consider Linux LVM snapshot. Is it possible to perform LVM snapshot between two volume group?
- If LVM snapshot can only be performed in a volume group, can I disband volume group : vg1, add all the physical volumes in vg1 to vg0. Basically I will have a single volume group : vg0 with 6 PV (2TB+2TB+1.4TB+2TB+2TB+1.4TB). After that, can I then create a LVM snapshot backup of the production database?
I think the above question is quite convoluted. I welcome any clarifications. Thanks in advance for all the gurus' help!