Not sure which RAID level to implement...

[mmp121]

I'm most likely going to pull the trigger on a NAS purchase in the next month. Now that most of my concerns of updated hardware (USB 3.0 & SATA 3.0) have been addressed by QNAP and the TS-x59 Pro II family.

Since this is my first NAS, and my first foray into the RAID world, I was hoping to levee the knowledge of those here who know much more than I do.

I have a fileserver using JBOD that serves upwards of 3.5TB of data (music / videos / photos / movie rips / docs / backups etc.) for my home network.

For my very important stuff (videos / photos / music / docs) I keep copies on external drives (500GB & 320GB).

I have my eye on the TS-659 Pro II (unless a TS-859 Pro II comes out before I buy ).

My question is to try to find a balance of speed & redundancy with my RAID implementation.

What would it be better for me to do assuming 3TB drives:

5 drive RAID-5 with (1 hot spare) 12TB space
6 drive RAID-6 with no hot spare 12TB space
6 drive RAID-10 with no hot spare 9TB space

My concern is (based on what I have read) that rebuilding the array if a drive goes down with RAID-5 can take a long time, and possibly lead to the entire array to fail.

Guidance is very much welcome!

 

[Gothri]

There is no one answer to these questions that is equally suited to everyone.

How much use will your storage array be seeing?
Is continuous, 24/7 up time critical, or can you tolerate downtime?
Is your data critical or mission critical?
What other back up solutions are you using?
How many concurrent accesses to your array do you anticipate?

In the end, based on all these considerations, you have to decide what your own tolerance for risk is and how much money and resources you have available to devote toward addressing that risk.

For many people using a NAS as a home file and media server, there is no imperative for continuous up time however the data is often irreplaceable if it includes memories-based media.

For home media / file server use maximum throughput is not as big an issue since there is not likley to be a high number of concurrent users, and a big upfront investment for the highest levels of throughput may prove to be a cash outlay that provides no additional value. For example, SATA 2.0 vs. 3.0 NAS's for home use: I'm not sure there's any current benefit that would be realized over the lifetime of your NAS. You'd want to know if there is likely to be one for your situation.

RAID-5 and 6 have comparable resync times. All that RAID-6 provides is some additional peace of mind in the case that an unrecoverable read error should occur while you are resync'ing after a single drive failure. You'll want to decide if this additional peace of mind justifies the cost of an additional drive and lower available space on your volume.

Enterprise-class drives are often much more expensive, but have longer warranties. Their error rates are typically certified to an additional order of magnitude. This matters if your NAS is operating in a high volume environment, but is probably unimportant if your NAS is operating in a low volume environment.

Everyone agrees that RAID is no substitute for back ups. Which ever NAS and array level you choose, if the data is critical and/or important, you still ought to devote resources to keeping it backed up.

For my own part, I am uninterested in arrays with a high number of drive bays. If the chance for a drive failure at any given moment is n, then in an 8-drive array your chance for a failure is 8*n. As the number of drives increases, your need for fault tolerance increases as well (and all that that entails). For home use, 4 to 6 drives ought to be more than plenty for most users.

 

[teknojnky]

6 drive RAID-6 with no hot spare 12TB space

This.

Why?

because even if you do raid 5 + hot spare, during the resync process your data is NOT PROTECTED.

raid-6 = raid 5 + hot spare that is already synced.

for example, lets take your raid 5 + hot spare and remove a disk to simulate a failure

- array is non-redundant
- hot spare automatically kicks in and starts a resyncronization
- a second disk fails (or is removed) during the resync
= data gone.

raid-6 (dual redundancy) in the same scenario;

- array is still redundant (its now essentially a raid-5)
- no hot spare, so you must manually replace the failed drive
- failed drive is replaced, and resyncronization starts (array is still redundant)
- a second disk fails during resync
- array is no longer redundant, but your data is still intact and the resync will bring it back to raid-5 single redundancy.


obviously if a 3rd drive failed with raid 6, then you would lose data, but the risks of that are much lower than with raid 5 + hot spare.

Essentially, you should highly consider raid 6 (dual redundancy) with any device of 5 or more bays.

most 4 bay devices don't support raid 6, although raid-10 would be pretty much equivelent if supported.