asked on
One RAID array or two RAID arrays
From performance standpoint, two array system makes more sense. Beyond that, what are the benefits of having two arrays?
If the server crashes and I have to resort to a complete system restore from Windows Server Backup, I can see single array system would take more time because it has to restore user data as well as OS.
I think about whether to set up OS and user data in one RAID or create two RAID arrays
on the same disks or different groups of disks?
if you split the disks with 2 different arrays then you could hit performance issues as both arrays are using the same physical drives
not disagreeing about separating OS from data, just need to know more about the physical server setup in terms of the number of disks you have
would personally create a raid 1 for the OS and raid 6 or 10 for data (depending on what it is being used for)
ASKER
@Seth
I meant the following when I mentioned two arrays:
I would put first 3 HDs (two on RAID 1 plus hot spare) for OS and the other 3 HDs (two on RAID 1 plus hot spare) for user data.
All 6 HDs would be handled by one RAID controller.
I agree with separating the OS from the data, but would approach it differently.
I'm assuming that the bare metal server will simply host VMs. If you are using a single array, set up the host to use the full array. The VMs (one for AD, the other for file/print/app serving) would exist on the host server. Set up a second VHDX file for the data for the file/print/app server.
If you wish to do two arrays, make the first one small, just to hold the host. 64G is likely to be plenty. Put the two VMs and the data VHDX on the second array.
With 6 physical drives, I'd consider setting up RAID 10 with 4 drives and use the last two as hot spares. Then continue with my suggestion above about using a single array.
ASKER
@CompProvSolv
“RAID 10 with 4 drives and use the last two as hot spares.” - makes sense if all 6 HDs are the same size. But I am getting three 480GB SSDs for OS and three 12TB SATA to store data.
ASKER
@andyalder
You are correct. I used wrong scenario for my question.
Let me try my question in a different way.
If I need to have 3TB of free disk space to store user data, would it be better to have
(1) Two 4TB HDs (on RAID1) and install server OS (c:\windows) and keep user data in the same logical disk (c:\userdata)
(2) On two 4TB HDs (on RAID1), create two logical disks - “c:\windows” on 1TB disk and “e:\userdata” on 3TB disk. Or
(3) Install OS on two 1TB HDs (on RAID1) - “c\windows” and keep user data on two 4TB HDs (on RAID1) - e:\userdata.
If the money is non-factor, what option would you prefer? Which option is the best for disaster recovery situation - like if you have to restore everything from windows server backup?
I'd install Server on the first array, using all of the 480G for it. I'd set up the DC and file/print/app server VMs on the 480G array as that should be plenty of space. Then I'd create a VHDX file on the 12T array for the data.
"If the money is non-factor, what option would you prefer?": I'd use NVMe devices for everything in RAID 10. One big array with hot spares, two VMs (DC and file/print/app) and a VHDX file for data. If your server won't boot from NVMe devices, then use four SSDs in RAID 10 with one or more hot spares and four NVMe devices in RAID 10 with one or more as hot spares. If money is really not a factor, then a server that can boot from NVMe devices would be preferred.
I'd always separate the data from the OS, & for that several arrays are ideal. It is also easier to backup.