PerpetuallyConfused
Cadet
- Joined
- Sep 4, 2020
- Messages
- 1
Hello, I'm new to the world of FreeNAS and building a NAS in general. Actually first and foremost I will not be using FreeNAS and ZFS on my first build because it's quite intimidating in terms of amount of knowledge uptake required and I will be using Linux and standard RAID with mdadm (which I am not an expert with in the first but until I make the jump in the future to FreeNAS it's all I'm comfortable with atm) so if my question is not allowed due to being off-topic just let me know and I will only ask FreeNAS related questions here...
I am planning on purchasing an 8-bay case such as the Silverstone CS381 or the U-NAS NSC810A. First off I don't see any reason not to fill up all the bays, so I will be using as many of the 8 drive slots as possible. I found some "triangles" while reading about RAID telling you to pick 2 of the 3: Fast, Cheap, Fault tolerant, and my choice is Fault tolerant first, then cheap, with speed being the least important.
I've narrowed down the possible RAID levels to only three: 6, 10, and 60.
Here's what I've picked up so far:
RAID 10: Fastest of the three, no parity calculations and quick rebuilding time, at best can tolerate 1 failure in each mirror, with a maximum 50% (four) disks failing as long as each mirror still has a remaining disk, otherwise 2 disks in the same mirror will bring it down, quite inefficient in terms of space - 50%, n/2 (but that's OK)
RAID6: Slower due to parity calculations, but RAID5 is no longer an option with large disks and chances of failure during rebuilding. Can tolerate any 2 disks failing. Minimum 4 disks. Actually a bit confused right now because I didn't know RAID 6 could scale upwards but apparently the limit is quite high (16, or 32, depending on sources I've found). It seems (uneducated guess) to me that a 4 disk RAID6 would be more safer than a, say, 8 disk RAID6, because a higher percentage of the space is used for parity, but I'm not sure. Anyhow obviously space utilization would be higher at n-2, so with 8 drives at max 75% usable space.
RAID 60: Basically a RAID0 of two RAID6's. I've read some sources saying it's better than RAID6 in every way (including rebuilding), and it's somewhere between RAID6 and RAID10. It can survive 2 failures in each RAID6 leg, so a total of any 4 disk failures in my case. So in some ways similar to RAID10 except RAID10 can only tolerate 4 specific disk failures, or at worse 2 will kill the array. Again quite inefficient at 50% space utilization like RAID10 at n/2
Also I suppose I could use less than 8 slots and make a smaller array but use some of the slots for a hot spare if necessary (please advise)
The biggest issue would be low levels of usable space as low as 50% for RAIDs 10 and 60 and since the cases only have 8 slots, upgrading would mean enlarging each drive in the array rather than adding more drives. But that is not too big of a problem with 10TB + drives being relatively affordable and hdd capacity growth in the future.
So my question is, for an 8 drive case, should I use RAID 6, RAID 10, or RAID60, and should hot spares be involved? I am leaning towards RAID 60.
I am planning on purchasing an 8-bay case such as the Silverstone CS381 or the U-NAS NSC810A. First off I don't see any reason not to fill up all the bays, so I will be using as many of the 8 drive slots as possible. I found some "triangles" while reading about RAID telling you to pick 2 of the 3: Fast, Cheap, Fault tolerant, and my choice is Fault tolerant first, then cheap, with speed being the least important.
I've narrowed down the possible RAID levels to only three: 6, 10, and 60.
Here's what I've picked up so far:
RAID 10: Fastest of the three, no parity calculations and quick rebuilding time, at best can tolerate 1 failure in each mirror, with a maximum 50% (four) disks failing as long as each mirror still has a remaining disk, otherwise 2 disks in the same mirror will bring it down, quite inefficient in terms of space - 50%, n/2 (but that's OK)
RAID6: Slower due to parity calculations, but RAID5 is no longer an option with large disks and chances of failure during rebuilding. Can tolerate any 2 disks failing. Minimum 4 disks. Actually a bit confused right now because I didn't know RAID 6 could scale upwards but apparently the limit is quite high (16, or 32, depending on sources I've found). It seems (uneducated guess) to me that a 4 disk RAID6 would be more safer than a, say, 8 disk RAID6, because a higher percentage of the space is used for parity, but I'm not sure. Anyhow obviously space utilization would be higher at n-2, so with 8 drives at max 75% usable space.
RAID 60: Basically a RAID0 of two RAID6's. I've read some sources saying it's better than RAID6 in every way (including rebuilding), and it's somewhere between RAID6 and RAID10. It can survive 2 failures in each RAID6 leg, so a total of any 4 disk failures in my case. So in some ways similar to RAID10 except RAID10 can only tolerate 4 specific disk failures, or at worse 2 will kill the array. Again quite inefficient at 50% space utilization like RAID10 at n/2
Also I suppose I could use less than 8 slots and make a smaller array but use some of the slots for a hot spare if necessary (please advise)
The biggest issue would be low levels of usable space as low as 50% for RAIDs 10 and 60 and since the cases only have 8 slots, upgrading would mean enlarging each drive in the array rather than adding more drives. But that is not too big of a problem with 10TB + drives being relatively affordable and hdd capacity growth in the future.
So my question is, for an 8 drive case, should I use RAID 6, RAID 10, or RAID60, and should hot spares be involved? I am leaning towards RAID 60.