Is storage spaces flexible enough for me?

Man of Honour
Joined
13 Nov 2009
Posts
11,635
Location
Northampton
I currently have a custom build home server based on a dual core and chip with an Intel nic running esxi, which then runs FreeNAS and a couple of other vms and currently have 2x1tb drives in a ZFS pool.

The plan is to run an 8 drive array eventually with 2 parity drives. However I'm at a point where my current storage needs are close to exceeding to storage space I have, but I haven't got the funds to drop on 8 drives and a controller card in one go.

I could go out and buy another drive to add to the pool but that presents me with two problems, firstly it means i still haven't got any redundancy and secondly once I have built my 8 drive array the 3+ disks in the pool cannot be added back to the RAIDZ2 array to increase capacity or the number of parity drives.
.
Is it possible with storage spaces to start with a JBOD type pool of drives then at a later date stick another drive in the system and add that as a parity disk or does it have the same limitation as ZFS whereby once the array is created as a JBOD I'm stuck with it being a JBOD array.
 
Potentially yes.

When you create a pool of disks you don't at that point specify how any data should be stored on them. It's only when you create a virtual disk that sits on your drives that you specify the type of data storage (parity etc.). So provided you have enough space somewhere to put your data temporarily, you could remove an existing virtual disk, add a drive to the pool, create a new virtual disk using a different data storage scheme and copy your data back.

None of that is automatic though ;)

Parity on Storage Spaces, even in R2, is quite slow. It would probably be fine if it's just for a media archive or similar low priority data, but I wouldn't use parity for anything too demanding.

Have you considered something like FlexRAID? It will take a number of drives with or without existing data on them and pool them, with optional disk(s) for parity. Ideal for media archives and possibly more flexible for your needs than Storage Spaces.

Just my 2p :)
 
I had a play earlier using VHDs and I can't do what I was hoping for.

The goal or this wad primarily to move away from virtualising the NAS part of the software while still being able to run a few pieces of windows specific software. So even as you say flexraid will achieve what I was hoping for in terms of disks I'm still going to have to virtualise the file storage.

EDIT : just took a quick look at flexraid and it appears it isn't quite what I though and will do what I wanted to achieve. Is flexraid going to be happy to take my existing drives that are ZFS formatted and holding data?

EDIT EDIT: More think etc. I honestly think my current setup is heading in the right direction. I'm simply not going to be able achieve all of what I want without some. From some reading flexraid has a lot of niggles and bugs which cause a lot of problems. I think ill stick with ESX, FreeNAS and server 2012 as it's setup now.

Change the motherboard to something that supports IOMMU and buy a dedicated SATA controller as intended and pass that through to FreeNAS
 
Last edited:
You can expand zfs pools. I started with 3x1tb = 2tb usable space and gradually replaced them all to 3x2tb. Once the last drive finished re 'slivering' it auto expanded to give me 4tb usable space
 
Last edited:
I know I can use ZFS in that way :)

I was hoping for a solution whereby I could add more physical drives.

I would like to run an 8 drive RAIDZ2 when the server is full populated but I can't afford £750 on drives a sata controller and a motherboard that allows hardware pass through.

At the moment I'm running two drives in a pool which is nearing being full so I'll add another drive to that pool.

I was hoping I could gradually build the number of physical disks to boost the amount of storage I have.

However I think I've found a solution with ZFS that achieves as close to possible as I would like by pooling two raidz1 arrays and a RAIDZ mirror.
 
Would I be right in thinking that this setup would give substantially better write speeds than the original planned 8 drive RAIDZ2 array, and also slightly more redundancy against drive failure.

Although as this is purely for backup and media storage disk performance isn't a huge issue.
 
Back
Top Bottom