Hello all, quick question about the new storage model 
Please excuse any ignorance on my part, I'm still learning about everything proxmox has to offer, and I'm merely a server administrator, not nearly smart enough to understand all the theory that goes behind putting something like this together!
First a little background:
We are running proxmox on a Dell PowerEdge 2950, 2x2.5ghz quad core, 32gb ram. We have a Dell MD1000 15 drive enclosure hooked up to a Dell PERC 6/e raid controller for the server. There are 15x 1TB SAS drives in the external enclosure, 14x in RAID6 array with 1x hotspare drive, yielding about 11tb usable space. The internal array is a PERC 6/i raid 10 array, yielding 6tb usable space (technically, it is 3x 2tb raid 1 arrays that I threw together in the same VG to make 6tb usable, did that to get around volume > 2tb issues). I installed proxmox 1.4b2 and it set up a 32gb swap, 100gb root, and the rest as vm storage.
We are trying to implement BackupPC for our department, and we are going to be backing up a few terabytes of data with it. So on to my question:
What would be the best way to handle the storage situation for BackupPC?
My initial thought was to use the internal array for most of the virtual machines, while using the external array mainly for BackupPC, giving it space to grow. I was concerned however that with 4kb block size, the maximum file size on an ext3 volume is around 2tb. This would mean that my image, if stored on an ext3 volume, couldn't grow past 2tb. This led me to think about using LVM inside the VM to concatenate multiple image files under 2tb into one large volume inside the VM, however I was worried about the management and maintenance issues this would introduce.
Then I noticed that when I created a new VG using the external array as a PV, that new VMs put on that storage were using raw image files and had /dev/ mappings for the images (I first created the VG using the external array as a PV, then I set up the storage via the web interface, choosing the new LVM VG as a base for new storage). Does that mean that image files placed on the external LVM'd storage will not be constrained by a 2tb limit because the storage is being used in a raw form (meaning not as image files being stored on a filesystem)?
I apologize in advance for wasting everyone's time if this is a silly question, I looked for a while for an answer, but I would like someone a whole lot smarter than me to tell me if this is feasible or accurate.
Thanks so much in advance!
Jordan Shinall
College of Education
University of Florida
Please excuse any ignorance on my part, I'm still learning about everything proxmox has to offer, and I'm merely a server administrator, not nearly smart enough to understand all the theory that goes behind putting something like this together!
First a little background:
We are running proxmox on a Dell PowerEdge 2950, 2x2.5ghz quad core, 32gb ram. We have a Dell MD1000 15 drive enclosure hooked up to a Dell PERC 6/e raid controller for the server. There are 15x 1TB SAS drives in the external enclosure, 14x in RAID6 array with 1x hotspare drive, yielding about 11tb usable space. The internal array is a PERC 6/i raid 10 array, yielding 6tb usable space (technically, it is 3x 2tb raid 1 arrays that I threw together in the same VG to make 6tb usable, did that to get around volume > 2tb issues). I installed proxmox 1.4b2 and it set up a 32gb swap, 100gb root, and the rest as vm storage.
We are trying to implement BackupPC for our department, and we are going to be backing up a few terabytes of data with it. So on to my question:
What would be the best way to handle the storage situation for BackupPC?
My initial thought was to use the internal array for most of the virtual machines, while using the external array mainly for BackupPC, giving it space to grow. I was concerned however that with 4kb block size, the maximum file size on an ext3 volume is around 2tb. This would mean that my image, if stored on an ext3 volume, couldn't grow past 2tb. This led me to think about using LVM inside the VM to concatenate multiple image files under 2tb into one large volume inside the VM, however I was worried about the management and maintenance issues this would introduce.
Then I noticed that when I created a new VG using the external array as a PV, that new VMs put on that storage were using raw image files and had /dev/ mappings for the images (I first created the VG using the external array as a PV, then I set up the storage via the web interface, choosing the new LVM VG as a base for new storage). Does that mean that image files placed on the external LVM'd storage will not be constrained by a 2tb limit because the storage is being used in a raw form (meaning not as image files being stored on a filesystem)?
I apologize in advance for wasting everyone's time if this is a silly question, I looked for a while for an answer, but I would like someone a whole lot smarter than me to tell me if this is feasible or accurate.
Thanks so much in advance!
Jordan Shinall
College of Education
University of Florida
