RAID question

[detjo]

I'm confused.

Just bought a new server with 8x300Gb SAS drives. I planned on setting up RAID 5 with all 8 disks, but I'm seeing comments from people that RAID 5 is a bad choice and that only RAID 1 or 10 is good.

 

Anyone care to explain?

[themightymrp]

Not that I've tried it but wouldn't RAID 6 be better (if your system supports it?). At least that can cope with 2 drive failures at once

[mrbios]

raid 10 is often the raid of choice, raid 5 is a bit pointless imo. If you need resilience above all else then raid 6 is the way to go, but in most circumstances 10 is good for performance and resilience.

[themightymrp]

Surely the compromise is that RAID 10 only gives you effective storage of 4 of your 8 drives though? With it being 4 drives striped together and then mirrored to the other 4? Or am I wrong. I should say it's been a while since I read up on my RAID - our servers are all RAID 5...

[tmcd35]

It's usually a question of speed, capacity and resilience - pick 2. RAID10 will give you speed and resilience, RAID5 will give you capacity and resilience, RAID0 will give you speed and capacity. Take your pick.

 

RAID10, you've got 1.2Tb space

RAID5, you've got 2.1Tb space

 

RAID6 is new and not always supported by all controllers. If it is then you get more resilence and less speed.

 

Horses for courses. The question is more what are you using the drives for than which RAID is best. Example, we have a 16x450Gb SAS storage array. I opted for RAID50 with 2 hot spares. Gives me the best ratios of speed, capacity and resilience in that particular set up. RAID10 would absolutely kill my available disk space, which wouldn't be acceptable on that server.

Edited February 10, 2014 by tmcd35

[Koldov]

Well, so far I have found this on WIki:

 

Unrecoverable read errors during rebuild[edit]

 

Unrecoverable read errors (URE) present as sector read failures. The unrecoverable bit-error (UBE) rate is typically specified at one bit in 1015 for enterprise class drives (SCSI, FC, SAS), and one bit in 1014 for desktop class drives (IDE/ATA/PATA, SATA). Increasing drive capacities and large RAID 5 redundancy groups have led to an increasing inability to successfully rebuild a RAID group after a drive failure because an unrecoverable sector is found on the remaining drives.[4][50] Parity schemes such as RAID 5 when rebuilding are particularly prone to the effects of UREs as they affect not only the sector where they occur but also reconstructed blocks using that sector for parity computation; typically an URE during a RAID 5 rebuild leads to a complete rebuild failure.[51]

 

Atomicity: including parity inconsistency due to system crashes[edit]

 

A system crash or other interruption of a write operation can result in states where the parity is inconsistent with the data due to non-atomicity of the write process, such that the parity cannot be used for recovery in the case of a disk failure (the so-called RAID 5 write hole - see below).[4]

 

This is a little understood and rarely mentioned failure mode for redundant storage systems that do not utilize transactional features. Database researcher Jim Gray wrote "Update in Place is a Poison Apple" during the early days of relational database commercialization.[58]

 

RAID write hole[edit]

 

The RAID write hole is a known data corruption issue in older and low-end RAIDs, caused by interrupted destaging of writes to disk.[59

 

RAID 5 in enterprise environments[edit]

 

RAID 5 is seriously affected by the general trends regarding array rebuild time and chance of failure during rebuild.[56]

 

In August 2012, Dell posted an advisory against the use of RAID 5 in any configuration and of RAID 50 with "Class 2 7200 RPM drives of 1 TB and higher capacity".[64]

 

With a RAID 6 array, using drives from multiple sources and manufacturers, it is possible to mitigate most of the problems associated with RAID 5. The larger the drive capacities and the larger the array size, the more important it becomes to choose RAID 6 instead of RAID 5.[65] RAID 10 also minimizes these problems.[52]

 

From here:

 

RAID - Wikipedia, the free encyclopedia

 

From that, I think it is mainly due to the re-build times and maybe if the parity bit is corrupt then it cannot be rebuilt.

 

The other RAID options you mentioned are basically mirrors so if one is lost the other is a perfect copy. I think however, in file servers or servers reading and especially writing multiple small files constantly, these RAID configurations add a lot to the processing overhead, writing all the data twice and halve your capacity.

 

I'm sure others here will give you a far better explanation, than my limited knowledge (and Googling) permits.

 

Kol.

 

Edit: Whoops, took too long in my reply... What the others said!

Edited February 10, 2014 by Koldov

[pcstru]

@Koldov - that's good info. Your google-fu is strong!

 

What no one has mentioned about Raid 10 is that in any configuration, the failure of two drives can wipe you out. So with 20 drives you might have 10 mirrors in that each individual drive has a mirrored copy, but when a drive fails, if the partner drive fails before you have the array rebuilt, you are right royally stuffed.

[themightymrp]

How about looking into Storage Pools with Server 2012?

 

Step-By-Step: Creating a Tiered Storage Space - Canadian IT Professionals - Site Home - TechNet Blogs

 

Supposedly gives the best of all features??

[PotNoodleTech]

It's a difficult decision.

 

After much testing of Raid5, 6, 10 I went Raid 50 here using 5 x banks of 3xSAS drives plus a hotpsare - I get some of the benefits of Raid 5 (capacity, speed) with some of the benefits of Raid 6 (multiple drive faliures, depending on which drive fails) with some of the benefits of Raid10 (fast rebuilds when degraded). Basically it's multiple raid 5 arrays of 3 drives, all wrapped up into a big raid 0. If a drive fails, only one x 3 drive raid5 array is degraded, so rebuild speed is super fast. Not all RAID cards support it though, and I always use SAS drives for reliability.

 

Start off with capacity (no point in considering raid 10 if it doesn't give you enough space)

Then move to reliability

Then consider speed (particularly if you need write speed).

 

People give Raid 5 a hard time, but it's a perfectly servicable and suitable method especially with low numbers of SAS drives. It's got a bad rep usually because it can be dangerous when used with large numnbers of unreliable desktop 4tb SATA drives with no hotspare (the typical configuration of low end nas/san's).

 

I've had raid 5 arrays of up to 8 drives work for 10 years with no problems.

[cogrady84]

I found myself in a similar situation recently when attempting to improve the resilience of our SAN before some serious P2V.

 

I was considering RAID-6, to improve redundancy, but after alot of research it still seems like RAID-5 is a viable solution. It has it's supposed problems, but in 12 years of the job, i've never had a single issue with RAID-5, i've had drive failures, but rebuilds have always been flawless. Keep your drives as similar as possible, drive firmware and controller firmware up to date and a solid backup solution and the choice of RAID strategy becomes much simpler.