What should i use for a RAID 5 setup

As you are finding out, there are multiple solutions available. If safety of your data is more of concern than a slight cost hit, then perhaps you should skip RAID 5 and consider RAID 10 (1+0). The cost factor is of course less storage space but you can have up to a two disk failure as opposed to RAID 5 (one disk). If you had several disks then there are more options but a limited amount (4 disks) is reasonable for the type of use required by you. As the costs of disks are going down, it becomes a viable option for home and small business users.

Software vs hardware RAID - Some enclosures handle the RAID for you with its own hardware/firmware solution. This is very handy and efficient (if it is a quality device) but then your RAID is vulnerable to failures of the enclosure and to restore would require a like enclosure. Software RAID as implied, means that your computer contains the software that creates and manages the RAID. I am unsure if the latest OSX still allows the creation of RAID but there are a couple of known software solutions out there that do quite well. If your enclosure fails, you can get another "dumb" enclosure and set up your drives in the proper order and it should continue to work as your computer should recognize the drives properly.

What comes to mind are makers such as OWC, Calldigit and Areca. The latter has some older TB1 4 drive unites that may be found on sites like Amazon. They come with impressive software RAID solution and is robust.

Last - other types of set ups that can be useful might include a 2 or more striped drives (RAID 0) for typical use and one additional drive that is equal or greater than the total volume of the stripe. This would reside in the same enclosure and the striped would be backed up from time to time (possibly use a scheduler). Thus, you use the RAID for work and the additional drive to back up the RAID. As for RAID 5, it works very well for most purposes but is based on simple swap out of a failed drive and the "restore" may or may not happen properly. I have had RAID 5 work for years, had to do a restore/rebuild when a drive died with no issues (though it took quite a bit of time). I was smart enough to know that the other drives could be problematic as they too were old and managed to replace all the drives over the following 2 months.

 

Don't do RAID5 with any disk > 1TB.
That's just a disaster waiting to happen.
The Areca-boxes offer RAID6. That's the way to go (if you don't want to do RAID10).
For RAID6, you need at least 6 disks (well, at least that's the minimum.

 

www.synology.com

[close thread]

 

If your data is important, then considering RAID6
I would recommend T12-S6.TB2, it offers RAID6 - 4 drives - to start, and you can add more drives to expand the raid as you need... Mean time you can use empty slots to hot swap any SATA/SSD to transfer data. like hot swap docking.

Since it's a 12 bays, If used 6TB HDD, at final set up you would have 72TB raw or 96TB if you use 8TB HDD

 

1st RAID5 requires 3x HDD minum, where RAID6 needs 4x HDD - not 6x HDD

2nd RAID5 only offers ONE HDD failed protection, meaning if another hdd failed during rebuild process, your data is toasted.
RAID6 offers TWO HDD failed protection, meaning if another hdd failed during rebuild your RAID volume is still OK and functioning.

 

Let's just say for the moment that R 6 is better than R 5 in terms of drive failure and rebuild challenges. This leaves R 6 vs R 10 (1+0).

With RAID 6, you get to enjoy the risk of up to 2 drives failing. The limitation is off set by having ANY 2 drives fail within your RAID and not any specific drive. With RAID 10, you could potentially have more drives than 2 fail (if the RAID is 6 or more drives) and still recover. The failure of drives is related to the "pairs" of drives creating a mirror. You can have only one of the drives within a pair fail. If you have 6 drives, you could have a failure on either side of the mirror as long as its not on both sides of the same set. This is based on a proper controller and not all controllers are equal.

1a 2a 3a
1b 2b 3b In this set up, you could have as example, 1a 2a and 3a all fail and still be running, or perhaps 1a, 2b, 3b where 1b, 2a and 3a failed. The reason is that the "stripe" still exists. If 1a and 1b fail, then the volume has failed.
So if the max number of drives failed, then the time between replace plus rebuild to full working state also becomes a risk.

Both RAID 6 and RAID 10 can have challenges with rebuilds but they are least likely as compared to RAID 5.

Speed - RAID 6 and RAID 10 can have similar read speed but on paper, RAID 10 is faster (in the real world, there are RAID 6 boxes that do have faster reads than other similar drive size RAID 10 boxes). As for writes, RAID 10 on paper is faster and in the real world, often echoes what is on paper.

Restoring a damaged RAID - not many people will challenge this as RAID 10 is faster at restoring than RAID 6 as there is limited calculations required. RAID 6 while taking longer than 10, may be faster than RAID 5 given that the process requires only 1 of the 2 parity locations be found and first found gets used (2searches vs 1search).

Just my take here -

If you have the need for speed, consider RAID 10, otherwise, take a slight performance hit and with proper planning, RAID 6 should be ideal for most purposes for home use. RAID 6 also offers more available space for a volumes than RAID 10 if that is a concern (given how cheap large volume drives are these days). For smaller files (such as photo files), RAID 6 is fine, if one is working with very large files such as video, they might find the difference of speed worthy of going to RAID 10.

I am sure there are people that disagree and this type of discussion will continue long after this thread is aged and old news.

Edit - If you have RAID 6 with 4 drives, you are getting in effect the use of two drives, this is akin to the same amount of storage (approximately) as 4 drives doing RAID 10. As such, the RAID 10 will offer better performance and you have two drives fail IF* they are not of the same RAID 1 mirror while the RAID 6 permits any 2 drives to fail. This is the trade off on a 4 drive RAID 6 vs 10.

 

Others have covered the types of RAID to choose so I'm not going there. However, consider why you want RAID. RAID is not a data backup strategy. It is for high availability. A failure (or other loss) of the RAID enclosure kills all your data. You still need appropriate backups.

I'm a real fan of Synology NAS devices. Easy to setup and administer. Easy to attach external drives for backup.

 

Really?

 

No. 4 disks is the minimum for RAID 6.

 

I know! Never fails to amaze me that people with such little knowledge can offer such definitive advice on a forum. There's nothing wrong with using RAID5 with disks > 1TB either, so long as you have an appropriate backup strategy, as we should all know 'redundancy' is not the same as 'backup'

 

True, you only need 4 disks.
But personally, I don't create RAID6-arrays with less than 6 disks.

 

Ah, right so your words of wisdom should have read "For RAID6, you need at least 6 disks (well, at least that's the minimum I would use). You can do it with just 4.

 

"Waste" or "Overhead" is bad enough with 6 disks anyway.
I should admit I only use it on servers - and then, mostly ZFS software-only RAIDz2, as it's called.

There's ZFS for OS X - but I have no idea how well it performs.
With increasing capacities in disks and to be able to create larger arrays, people are actually transitioning to RAIDz3 (three parity-disks).
People should not underestimate the time it takes to rebuild a large array that is near or at capacity. Even more so, if you want to be able to actually use it while it's rebuilding.

Of course, it's a free country and you can use whichever RAID-level you like.
Just don't complain when it all explodes in your face ;-)


People might want to read this very interesting blog-post from one of the creators of ZFS:
http://blog.delphix.com/matt/2014/06/06/zfs-stripe-width/

 

For the backup drive, there is no need to go with any kind of RAID if you can buy a single disk of the required size. Your best setup would be to..

1. Buy a large disk (4TB) and use it with Time Machine. TM works well and is nearly fool proof.
2. You need a redundant backup. The simples is to just buy a second drive like the one above and TM will keep both up to date
3. In addition to the above you need an off site copy of your data. The data needs to remain off site even while it is being backed up. The simplest option is to subscribe to an on-line service. There are many

Think about what are the most common ways to loose data, RAID does not address them.
1) A file is silently corrupted. You have a good backup copy. So far so good. But then you do this "copy to try backup drive" and over write the only good copy of that file. You will not know your done this until you want there look at the photo. This is why Time Machine is good. It save all the old versions of your files and does not over write them (until it runs out of room.)
2) Your house burns down or some one breaks in an takes all your computer stuff or lightening hits a power pole down the block and everything that is plugged into an AC outlet is fried. You need two copies of the data that is not on-line at your house/office

What I do
1) Keep a large Time Machine drive connected,
2) then periodically I make copies to bare SATA drives to keep in a fire safe in another room
3) Use an on-line service, they continuously keep up to date with changes as I make them.

The rule of thumb should be that no matter what kind of disaster strikes you need to always have your data AND a backup.

About RAID. There are a lot more then just the common RAID 1, 2, 3, 4, 5, 6, 10. For example Drobo has their own system which is really pretty good as it will take best advantage of however many drives you give it. But the "best" might be ZFS. First off ZFS is a copy on write system which means it does not over write old data (until forced to) it also has very robust error detection that keeps checksums all the way through the entire write/read process. It also has RAID-like redundancy. It is one layer, not a RAID with a file system on top. It is also free and open source on BSD. It can be VERY fast too. The down side is the hardware requirement. Google "FreeNAS"