Currently raid 10, should i be using raid5?

[Hitman36]

Hi,

I currently have a 453pro with 4x WD RED 3tb nas drives - raid 10 configured but unsure if I should have chosen raid 5 but it seems a lot of users have trouble with R5 and decided to go R10, did I make the right choice, not too bothered by the less storage with R10 but I am interested in having good failover?

I use surveillance station, file station only at the moment and have my Bluray .iso backups, other .MKV movies I playback through emby/mediabrowser classic via WMC (Win8.1x64) along with my many installer programs/photo's SW tools etc..backed up (i have installed/use on my main PC) in the personal folder.

If I should go with R5 can I migrate/convert R10 - R5 without losing data?

Thanks.

[dsmithdewarcom]

not too bothered by the less storage with R10 but I am interested in having good failover?

RAID 10 has better failover in a a 4-disk configuration, faster recovery in case of a disk failure, and better performance.

As you mentioned, the trade-off is capacity. If you can live with that, you made the best choice.

[Hitman36]

Thanks for the confirmation, i'll stay with raid 10.

[pwilson]

Thanks for the confirmation, i'll stay with raid 10.

It's about Risk Management for me. I am more than satisfied with RAID5 performance here, and I like the extra space. RAID10 is definitely safer (2 drive redundancy) vs only single drive redundancy with RAID5.

[P3R]

Only RAID 6 offer true 2 drive redundancy (where any two disks in a volume can fail at the same time and still recover) and therefore offer better fault tolerance than RAID 10 in a 4-disk array.

• RAID 5 - 1 disk redundancy, 3/4 of disks usable for data storage, good read performance, fair write performance.
• RAID 10 - 1.66 disk redundancy (actually 1 or 2 disks depending on which two disks fail concurrently), 1/2 of disks usable for data storage, fair read performance, best write performance.
• RAID 6 - 2 disk redundancy, 1/2 of disks usable for data storage, best read performance, worst write performance.

[dsmithdewarcom]

RAID 6 carries a high risk of an unrecoverable (meaning fatal) read error during a rebuild. Also rebuild time for RAID 6 is much longer due to parity calculations. RAID 10 is recommended in enterprise applications, for best balance of performance and redundancy. For some applications, RAID 5 or 6 may make more sense.

[P3R]

RAID 6 carries a high risk of an unrecoverable (meaning fatal) read error during a rebuild.

With RAID 6, 3 disks will be read but you still have a one disk redundancy so one of the 3 disks can fail and the rebuild will still complete.
With RAID 10 you're totally dependent on a single disk (with no disk redundancy left) being without errors.

Search for a RAID risk calculator and compare the MTTDL (Mean Time To Data Loss) between RAID 6 and RAID 10. Prepare for a surprise...

Also rebuild time for RAID 6 is much longer due to parity calculations.

True but some more hours isn't that critical when you still have disk redundancy left. Especially not in a SMB/home user application where high performance during a rebuild is normally not a requirement.

RAID 10 is recommended in enterprise applications, for best balance of performance and redundancy.

Yes enterprise applications require enterprise solutions. Enterprise solutions however aren't necessarily the best in SMB and home user applications.

The reason for RAID 10 being favoured in storage intensive (often database) enterprise applications is mainly performance, especially write performance that is the Achilles heel of RAID 6.

For home users there is much write only once (and then the bottleneck is normally ripping and/or download speed anyway) and read many times with streaming media applications.

In SMB, with clients using mixed office applications, it isn't critical if a write take 2 or even 5 seconds more as that usually happens in the background anyway.

For some applications, RAID 5 or 6 may make more sense.

Yes, for file servers, especially in SMB and home applications, they very often do. Clueless people think the higher the RAID number, the better it is but that's not true. Different RAID levels all have their pros and cons.

If the OP is not very, very performance oriented (and he also have a very nice and powerful NAS that will handle RAID 6 well), I'd think that RAID 6 would be a better choice than RAID 10 given the initial request for very good fault tolerance. But only the OP can decide if the improved fault tolerance is worth the hassle of a reinstallation, as I don't think there is a migration path from RAID 10 to RAID 6 in the SMB firmware either (my older model doesn't run SMB firmware so I haven't bothered to look at the details of it yet).

[storageman]

I have tested RAID 10 against RAID 5/6 on a few different models and RAID 10 doesn't make a hugh difference to read/write speed across 4 disks.
RAID 10 only has big difference when looking at large number of bays, say 8-12 upwards.

[Hitman36]

Thanks guy's for the extra information, i'm now interested in Raid 6 since my storage in now, surprise surprise running low :/

Problem i'm faced with is migration and the potential of data loss and/or getting my stuff off onto other temp storage.

I do have physical hard drive backups of my most important data but these are almost full and nowhere near the capacity I need for a full backup, hmm some thinking to do.

Thanks again.

[comas17]

With 4 disks, the available space with RAID10 and RAID 6 is the same
You will not have more disk space available migrating to RAID6

[mungbean]

RAID5 is getting a less viable option as hard drive sizes increase. Reading articles that crunch the numbers for you realistically only enterprise level drives are suitable for 2-3TB drives in RAID5. 4TB+ drives is not really advised for somewhat critical data.
RAID6 with 4 drives is usually slower than RAID10 with only very slight improvement in redundancy.
RAID10 is a good option for decent redundancy and performance but is more costly in terms of storage capacity lost.

[P3R]

RAID5 is getting a less viable option as hard drive sizes increase.

True but the OP only have 4*3 TB disks.

Reading articles that crunch the numbers for you realistically only enterprise level drives are suitable for 2-3TB drives in RAID5.

Ahh, you mean the old internet myth that started with this article and that uncritical and less knowledgeable persons still think is the truth? Already the heading of the article have prooved to be extremely incorrect and then there's a major flaw when the writer can't read disk data sheets correctly. Yet it is repeated over and over...

4*3 TB in RAID 5 is since 2011 a very common configuration and as long as staying away from known incompatible disks like Seagate DM/DL and WD Green, they are reliable. Of course RAID 6 and RAID 10 have better disk redundancy and are the only choices in mission-critical applications, when maximum availability is crucial. But for home and SMB servers RAID 5 is still by far the most used configuration. I'd say that today even 4*4 TB RAID 5 have proven to be reliable enough for home/SMB use with standard NAS or good (like HGST) desktop disks, with a URE=10^14.

4TB+ drives is not really advised for somewhat critical data.

Maybe but the OP doesn't have that...

With 5 and 6 TB disks we'll have to wait a few years to see if RAID 5 is still reliable enough. I don't know but I wouldn't count them out already though, and look stupid afterwards like Robin Harris...

RAID6 with 4 drives is usually slower than RAID10 with only very slight improvement in redundancy.

I'd say that with 4 disks in a modern NAS, RAID 6 is only very slightly slower and only when writing than RAID 10. RAID 6 gives true 2 disk redundancy while RAID 10 only give a 1.66 disk redundancy. Actually it's of course either a 1 or 2 disk redundancy but after the first disk have failed there's a 1/3 risk that the second disk failure is catastrophic. I don't like those odds in a mission-critical application.

RAID10 is a good option for decent redundancy and performance but is more costly in terms of storage capacity lost.

It's a good option for very write intensive enterprise applications and when using many spindles.

For the smaller NAS models (8 disks or less) normally used in home and SMB applications, RAID 6 is normally a better choice than RAID 10.

Since maximum storage is often the absolute most important factor for home and SMB users, RAID 5 is also a good compromise in 4 bay models. With 5 bays and more, especially with 5 TB and larger disks, I would normally recommend RAID 6. In 6 bay and larger NASes, RAID 10 may also be an alternative to some with specific needs.

[Hitman36]

Thanks again guy's very informative but shows i'm in a catch 22 situation :/

Real solution then is get bigger disks but I've only just had 4*3 running a few weeks, I had a feeling I should have got 4tb drives to start with but took me a few weeks to get 4 3's

I'll make do until the 4tb prices come down.

Thanks!

[mungbean]

Ahh, you mean the old internet myth that started with this article

Nah not that one. I saw a published article with real maths titles something along the lines of "No redundancy is more reliable". Concluded with you are likely to get errors during the rebuild process of RAID5 on low end drives 4TB +.
I had a couple of rebuilds on RAID5 arrays very successfully on 500gb-1000gb drives. Very time consuming.

On another note to solve your storage problem:
Toshiba had a special on their enterprise level drives recently and may still be going on approx 180 USD for 5TB (125 USD for desktop level). Not a lot of information floating around on them yet. They appear similar to HGST drives when I looked at them physically. I remember reading somewhere that Toshiba acquired HGST 3.5" manufacture via a deal with WD. So they should be solid drives.

[P3R]

I saw a published article with real maths titles something along the lines of "No redundancy is more reliable".

Please, please, please give us a link to that article. When I google "No redundancy is more reliable", the only relevant hit I get is this. Could it be that one?