Hello from The Pacific Northwest. I'm looking to build a FreeNAS server. I'm no stranger to servers in general but this would be my first FreeNAS server.
I have some (hopefully) easy-to-answer questions about ZFS. I don't know if this subforum or the Help & Support subforum is better for that . . . I'll try here first.
I've read a fair bit about ZFS, and think I have a grasp of the concepts, but am seeking clarification on a few points.
My first question involves the pool/vdev/disk hierarchy. My understanding is that a vdev is (typically) made up of several physical disks, then one or more vdevs comprise a pool. From browsing the FreeNAS web interface it looks like the term Volume is used to refer to a pool ('zpool' although according to the Terminology and Abbreviations Primer 'pool' is preferred).
I'm trying to compare/contrast an approach where several vdevs comprise a single pool to one in which each vdev belongs to its own pool (thus yielding multiple pools as opposed to just one). Here is what I'm seeing as the tradeoffs between the two approaches.
single pool with mutiple vdevs
Pros
- improved throughput due to striping across the vdevs
- content is automatically distrbuted across the vdevs
- any spares are shared by all vdevs
Cons
- loss of any single vdev in the pool results in loss of the entire pool
multiple pools with one vdev per pool
Pros
- loss of a single vdev does not cause a loss of the other vdevs/pools
Cons
- throughput is reduced due to lack of striping across multiple vdevs
- content must be read/written to a specific pool rather than to just the one
- no way to share spares across multiple pools
Have I at least gotten those Pros/Cons right, and what others might there be that I have yet to see?
I actually have a 'test server' built in the single-pool configuration right now (16x4TB drives) and am rsyncing to it from my current primary server, just to see how well it works as a backup server.
Thanks for your help!
I have some (hopefully) easy-to-answer questions about ZFS. I don't know if this subforum or the Help & Support subforum is better for that . . . I'll try here first.
I've read a fair bit about ZFS, and think I have a grasp of the concepts, but am seeking clarification on a few points.
My first question involves the pool/vdev/disk hierarchy. My understanding is that a vdev is (typically) made up of several physical disks, then one or more vdevs comprise a pool. From browsing the FreeNAS web interface it looks like the term Volume is used to refer to a pool ('zpool' although according to the Terminology and Abbreviations Primer 'pool' is preferred).
I'm trying to compare/contrast an approach where several vdevs comprise a single pool to one in which each vdev belongs to its own pool (thus yielding multiple pools as opposed to just one). Here is what I'm seeing as the tradeoffs between the two approaches.
single pool with mutiple vdevs
Pros
- improved throughput due to striping across the vdevs
- content is automatically distrbuted across the vdevs
- any spares are shared by all vdevs
Cons
- loss of any single vdev in the pool results in loss of the entire pool
multiple pools with one vdev per pool
Pros
- loss of a single vdev does not cause a loss of the other vdevs/pools
Cons
- throughput is reduced due to lack of striping across multiple vdevs
- content must be read/written to a specific pool rather than to just the one
- no way to share spares across multiple pools
Have I at least gotten those Pros/Cons right, and what others might there be that I have yet to see?
I actually have a 'test server' built in the single-pool configuration right now (16x4TB drives) and am rsyncing to it from my current primary server, just to see how well it works as a backup server.
Thanks for your help!
