Raid 5 Configuration

It is called a "hot-spare" ... a non-member disk dedicated to automatically rebuilding in the place of a failed disk.

You said you are using 4 disks in the VD/array currently, want to add the hot-spare as a member disk to expand your array, then add another hot-spare later, right?  You can do this IF you are running a RAID 5 or 6.  You cannot do this IF you are running a RAID 10.

You can add/remove a hot-spare non-destructively, while live, from OMSA (management software installed in the OS).

You can expand a RAID 5/6 by adding a disk non-destructively, while live, from OMSA (management software installed in the OS).

You can ONLY expand a RAID 10 from 4 disks to 6 disks by backup/delete/recreate/restore.

Thats great!  Tnx for the informative response.....

Regards...

Actually, revisiting the perc h700 config utility, it appears that all 5 drives are accounted for, drive 1&2 - 2x 1TB (931 GB actually) = 1843 GB VD, drives 3&4 - 2x 1TB (931 GB actually) = 1843 GB VD, and drive 5 38GB VD.  So in this config the 5th physical drive is underused.  

Is there a golden rule to how big a VD should be?  The server will be a file server so is it advisable to have 1 VD or is this unstable or less prudent than having 2 or more?

Tnx...

A few notes:

- a bootdisk can only be larger than 2TB (2048GB) if you change the bios to UEFI

- not all operating systems support booting to UEFI, but the more common ones that do are: VMware ESXi 5 (not 4.x), Windows Server 2008 x64 and 2008 R2, RHEL 6 (don't believe RHEL 5 does, but I'm not sure)

- if you split your raid 5/6 into 2 virtual disks (e.g. bootdisk and datadisk), you won't be able to expand the raid set with more drives

If the arrangement you posted is correct, then I might advise you to wipe everything out and start over, as it is risky and sub-optimal ... maybe you can confirm if this is actually true:

If you have two 1TB disks in the same VD and the size of that VD is 1843GB, then you have them in a RAID 0 ... RAID 0 has NO redundancy - one read/write error means data loss and if one disk fails, then everything on that VD is gone.

Two RAID 1's will give you only half the space, but will be redundant to help protect your data/server availability.
One RAID 5 will give you redundancy, and will also give you more usable disk space.

RAID 0's:  
VD0 - 1843GB, no redundancy;
VD1 - 1843GB, no redundancy;
VD2 - 38GB, no redundancy, majority of disk space wasted.

Two RAID 1's:  
VD0 - 921GB, redundant;
VD1 - 921GB, redundant;
VD2 - 921GB (RAID 0 for single disk), no redundancy

One RAID 5 (UEFI):  
VD0 - 3684GB, redundant; must use 64-bit OS installed on GPT-formatted disk

Two RAID 5 (BIOS):  
VD0 - 100GB (can be any size you want for OS), redundant;
VD1 - 3584GB, redundant, must be converted to GPT for OS to use full amount; cannot use "online capacity expansion" to expand VD sizes by adding disks

Tnx for the reply.  I'll use W2008 stnd R2 x64 and have a Raid 1 config for the OS.  I ask b/c hardware wise I understood there was a non-official guideline not to use larger than 1TB drives for servers...might be an old convention.  I wanted to check if there is something like a preferred size limit w/ VD drive setups.  I'll assume that there isn't and split the array to make two VD's.

Tnx...

I should have waited just a minute :)

The "conventions" change as hardware changes ... 2TB disks are fully supported on Dell's newer servers (not cheap consumer/desktop disks, mind you), so there is no reason not to.

With 2008R2, you could install in UEFI mode, IF you had wanted to lump them all into a single RAID 5 that exceeded the 2TB boot disk limit Dev Mgr was talking about ... but since you plan to install the OS on a RAID 1, that size limit isn't relevant.  On modern servers and OS's, the OS limits on boot disks is really the only consideration for the VD size.