Disk layout for 8 bay NAS

Soldato
Joined
17 Nov 2007
Posts
3,332
Hey All,

Just wondering what disk layouts people would go with for an 8 bay NAS.

Using WD REDs 4TB maybe something like this :-

1-2 : raid 1 mirror
3-4 : raid 1 mirror
5-7 : raid 5
8 : hot spare

or

1-2 : raid 1 mirror
3-4 : raid 1 mirror
5-8 : raid 5

Interesting to see what others would do and reasons why.
 
I don't think RAID5 is a good idea on big drives due to rebuild times and the risk of losing another disk during a rebuild. RAID6 increases fault tolerance by adding additional parity.

What about a RAID10 array? This will stripe over 4 disks and mirror to the other 4. Performance should be better than 5/6 and rebuild times much shorter as it's less computationally expensive on the CPU.
 
Raid 6 if capacity needed otherwise Raid 10.
No hot spare (and if you were considering Raid 5 then it's bad practice anyway).

This assumes it's a nas you can access easily to change disks (i.e. not in a remote location) and has decent alerts e.g. Synology or similar.
 
suprised no ones asked this already, what data is being stored on it and how important is it too you? also what capacitiy of data do you have to store? The raid type should be chosen around your requirements / usage scenario.
 
suprised no ones asked this already, what data is being stored on it and how important is it too you? also what capacitiy of data do you have to store? The raid type should be chosen around your requirements / usage scenario.

Almost any option is better than the OP's layouts though:


Using WD REDs 4TB maybe something like this :-

1-2 : raid 1 mirror
3-4 : raid 1 mirror
5-7 : raid 5
8 : hot spare

or

1-2 : raid 1 mirror
3-4 : raid 1 mirror
5-8 : raid 5

Option 1 gives 16TB of space, Option 2 gives 20TB.

RAID6 across all 8 disks would give 24TB, RAID10 only 16TB but with higher performance and better redundancy.
 
Are you using 4TB drives because you already have them?

Buying an eight bay NAS and then fully populating it on day one seems odd.

How much capacity do you need for now and the immediate future?
 
Almost any option is better than the OP's layouts though:




Option 1 gives 16TB of space, Option 2 gives 20TB.

RAID6 across all 8 disks would give 24TB, RAID10 only 16TB but with higher performance and better redundancy.


I'm interested how RAID10 offers better redundancy than RAID6.
If you lost a disk from each stripe on RAID10 would it still function ?
and I assume if the answer to that is yes, then if you lost "disk 2" from each stripe then you would lose the entire array ?

Personally I use RAID-Z2, I am not even sure if my NAS offers RAID10 (NAS4Free)
 
I'm interested how RAID10 offers better redundancy than RAID6.
If you lost a disk from each stripe on RAID10 would it still function ?
and I assume if the answer to that is yes, then if you lost "disk 2" from each stripe then you would lose the entire array ?

Personally I use RAID-Z2, I am not even sure if my NAS offers RAID10 (NAS4Free)

"Better" was probably the wrong word:

In an 8 drive RAID 10, best case you can survive up to 4 disk failures, providing they are not both from the same mirror pair.
Worst case is that if both disks within the same pair would be a loss of the array.

Whereas in RAID6/RAID-Z2, ANY two drives can fail without data loss.

The big difference however is that rebuilding a RAID10 setup only requires a single drive to be read, and requires no complex parity calculations to be done, so minimises both the risk window (as the rebuild is quicker), and the risk to other drives (e.g. increased load when reading information from all remaining drives to reconstruct the missing drive - which can lead to another drive failure, especially in RAID5 - where it would cause the loss of the array)
 
"Better" was probably the wrong word:

In an 8 drive RAID 10, best case you can survive up to 4 disk failures, providing they are not both from the same mirror pair.
Worst case is that if both disks within the same pair would be a loss of the array.

Whereas in RAID6/RAID-Z2, ANY two drives can fail without data loss.

The big difference however is that rebuilding a RAID10 setup only requires a single drive to be read, and requires no complex parity calculations to be done, so minimises both the risk window (as the rebuild is quicker), and the risk to other drives (e.g. increased load when reading information from all remaining drives to reconstruct the missing drive - which can lead to another drive failure, especially in RAID5 - where it would cause the loss of the array)
Thanks for explaining, I thought RAID 10 was handled slightly differently, ie 2 mirrored pairs of 4 drives, so if you lost a drive from each stripe you lost the array, but I guess that wouldn't make any sense.

I made the change to RAIDz-2 when I read about the stress it puts on the system doing a rebuild/resilver.

So now I still have the stress, but reduced the risk of total failure.


I have been looking in to online storage options as backup but not having much look with my data being on a NAS, at least not cheaply anyway.
 
I'm interested how RAID10 offers better redundancy than RAID6.
If you lost a disk from each stripe on RAID10 would it still function ?
and I assume if the answer to that is yes, then if you lost "disk 2" from each stripe then you would lose the entire array ?

Personally I use RAID-Z2, I am not even sure if my NAS offers RAID10 (NAS4Free)
Yes, it does although it's not named as such. You just create mirrored pairs of drives and add them to a drive pool. It's more flexible than RAID10 as the mirrored pairs don't have to be the same size as each-other.
 
Another option would be this :-

1-2 : raid 1 mirror
3-8 : raid 6

The reason being QNAP creates a bunch of defaults and from memory also stores upgraded firmware and apps on the initial volume, this would be the raid 1 mirror. The raid 6 set could them be created as an encrypted volume that would require unlocking once the NAS has booted and could store all documents etc.

Thoughts?
 
So with all the information above what about a layout like this :-

1-4 : raid 10
5-8 : raid 6

Using 4 disks for raid6 doesn't gain you anything over raid10 (as you lose 2 drives to parity).


If you need absolute reliability and minimal risk then 6 disk raid10 + hot spare (leaving 1 bay free)
8 disk raid 10 for performance and good reliability (but increased risk as no spare)
8 disk raid 6 for maximum capacity and good reliability (raid 5 and raid 6 you are better with no hot spare as you generally don't want an automatic rebuild that could cause a further failure)
 
OK just been checking on the QNAP configuration and you create your storage pool, ie raid set, then create your volumes on top.

Therefore could create a single raid 10 or 6 storage pool and have multiple volumes, the first smaller default volume unencrypted with a subsequent volumes either encrypted or not as needed.

The QNAP also support M.2 drives for read/write or just read caching.

EDIT: with the caching enabled I am thinking raid 6.
 
Using 4 disks for raid6 doesn't gain you anything over raid10 (as you lose 2 drives to parity).


If you need absolute reliability and minimal risk then 6 disk raid10 + hot spare (leaving 1 bay free)
8 disk raid 10 for performance and good reliability (but increased risk as no spare)
8 disk raid 6 for maximum capacity and good reliability (raid 5 and raid 6 you are better with no hot spare as you generally don't want an automatic rebuild that could cause a further failure)
How about small SSD for system, 6 drive RAID 10 plus hot spare?
 
Ultimately It all depends on the requirements.

Since you're using it for general office storage then you don't really need the speed advantage of a RAID10 array.. ALSO, as others have said, RAID 10 is NOT more secure than RAID6, don't buy into the hype :) You can loose 2 drives in RAID1- and loose everything, in RAID6 you can loose 2 drives safely without any data loss.

Yes, the rebuild time is a lot more, and you are at risk when rebuilding, however the probability of loosing another drive is slim (Even with the extra load a RAID6 puts on the disks).

Also, if you are looking at encryption, remember that this will have a large impact on the performance. Most QNAP devices (and Synology), don't have dedicated encryption processors so the main CPU will be doing the work, this means that throughput will drop significantly for any encrypted volumes.

Also, please remember that RAID doesn't protect data. It protects UPTIME. RAIDs ensure that in the event of a disk failure, you keep running. BACKUPS protect data. RAID is NOT A BACKUP SOLUTION

Stolen from another site:

The probability of a dual-member RAID1 outright dying is the probability of any given drive dying, squared: P(R1) = P(drive)^2.

So with the 5% failure rate, you end up with:

P(R1) = 0.005^2 = 0.0025 = 0.25%

(Here's where I'm not sure about the math---it makes sense, but that doesn't mean anything)

So the chances of losing any given R1 member are 0.25% per year, but you've got several of them striped together, and if any of them die, your array is dead. so you need to plug the P(R1) number into the R0 failure math: P(R10) = 1 - (1 - P(R1))^(n_R1). Lets assume you've got 8 drives total (and don't replace them when they fail) so a stripe across 4 R1s:

P(R10) = 1 - (1 - 0.0025)^4 ~= 0.99%

So you've got about a 1% chance of losing a RAID10 that's got 8 drives with a 5% chance of failure per-year.

Simplistically, the probability of RAID6 failing is the probability of any given drive dying, cubed (since you have to lose three drives for it to fail), e.g.:

P(drive)^3 = (5%)^3 = 0.0125%

So for the "multiple drives dying" scenario, RAID6 is about 1/80 as likely to fail as RAID10 with the same drives.
 
Back
Top Bottom