tips on raid ? :)

[mrbulle]

Well.. I've built my "server" .. it will run windows 7 and just have a home-group and shared drives/folders so everything i download from my PC and HTPC goes to it.. it has one 3tb drive that i will use to download random stuff to.. It has also 

Three 2tb WD green that i am considering doing raid on.. so the question.. which raid? (on the three drives) i want it to be a bit safe.. like if one drive goes then i can pop in a new one and save the data.. so raid 5 or what? i am just using the MB raid solution.. OS is on a 64 ssd. 

 

comments  is appreciated good folks  

[antoainb]

Well.. I've built my "server" .. it will run windows 7 and just have a home-group and shared drives/folders so everything i download from my PC and HTPC goes to it.. it has one 3tb drive that i will use to download random stuff to.. It has also 

Three 2tb WD green that i am considering doing raid on.. so the question.. which raid? (on the three drives) i want it to be a bit safe.. like if one drive goes then i can pop in a new one and save the data.. so raid 5 or what? i am just using the MB raid solution.. OS is on a 64 ssd. 

 

comments  is appreciated good folks

 

5, you won' get amazing speeds, but it's the best solution, since you have 3 drives.

[Zeke]

Yeah, RAID5 is basically the only somewhat safe option utilizing three drives. You could throw one of them out of RAID and go for mirror, or throw in one more and get RAID10, I guess...

 

Although rebuilding RAID5 array in case of a disk failure can be long (you're looking at about 12 hours here, might be even more, depends on hardware/software, less with dedicated RAID card, more with onboard RAID) and it really stresses the other hard drives during the rebuild - occasionally that might turn out really bad.

[mrbulle]

Yeah, RAID5 is basically the only somewhat safe option utilizing three drives. You could throw one of them out of RAID and go for mirror, or throw in one more and get RAID10, I guess...

 

Although rebuilding RAID5 array in case of a disk failure can be long (might take almost half a day or even more, depends on hardware/software, less with dedicated RAID card, more with onboard RAID) and it really stresses the other hard drives during the rebuild - occasionally that might turn out really bad.

 

cheers.. but another question then.. raid 10 vs raid 5.. is it like "better" or "safer"? does it stress the other drives less if one goes then? 

[mrbulle]

cause i can get antoher drive if its "worth" it.. i just have the three wd's lying around now......

[Zeke]

cheers.. but another question then.. raid 10 vs raid 5.. is it like "better" or "safer"? does it stress the other drives less if one goes then? 

With RAID10 you're getting 50% space efficiency, with RAID5 (three disks) - about 66%. Both allow you to lose one drive (you might be able to use RAID10 if two certain drives fail and if you have a good RAID controller). Rebuilding RAID10 is faster than RAID5 (you basically need the time it takes to copy the volume of a single drive). RAID10 is about 1.33x faster than 3-disk RAID5, but at the same time it's 1.33x more likely to lose a drive (since you have one more in RAID10. RAID10 is less CPU demanding (if you aren't using a dedicated controller) and handles random writes better than RAID5.

 

A little quote from here

Hitachi does not recommend using RAID 6 drives with high performance applications where extreme random writes are being performed. In some cases, the use of RAID 1 or RAID 1+0 is essential. 

From a performance standpoint, the performance of RAID 5 and RAID 6 is pretty similar when we talk about Random Read, Sequential Read and Sequential Write workloads. There is added penalty when we talk about Random Write workloads, that is because of the two dimensional parity. Compared to RAID 5, RAID 6 takes a 33% Performance hit on Hitachi with Random Write workloads. 

 

My opinion - go for RAID10 if you are using an onboard controller and able to get a fourth drive. Chances are you'll never ever ever experience an HDD fail during the rebuild of RAID5, but I'm kinda being paranoid here. RAID10 and 3-disk RAID5 have approximately the same amount of useful space, 4-disk RAID5 has a little more.

 

To sum up:

RAID10 - fast, easy to setup, less CPU demanding, easy to rebuild < this one is great overall, but you lose 50% of capacity

3-disk RAID5 - slower, but more storage efficiency, smaller chance to lose a drive (as there are only three instead of four), but more CPU-demanding and might cause more problems when rebuilding. < this one if you want to save money on HDDs

4-disk RAID5 - even slower, even more storage efficiency (75%), rebuilds even longer than 3-disk RAID5. < this one if you need more space and also a masochist

3-disk RAID2 - RUN AWAY.

[mrbulle]

With RAID10 you're getting 50% space efficiency, with RAID5 (three disks) - about 66%. Both allow you to lose one drive (you might be able to use RAID10 if two certain drives fail and if you have a good RAID controller). Rebuilding RAID10 is faster than RAID5 (you basically need the time it takes to copy the volume of a single drive). RAID10 is about 1.33x faster than 3-disk RAID5, but at the same time it's 1.33x more likely to lose a drive (since you have one more in RAID10. RAID10 is less CPU demanding (if you aren't using a dedicated controller) and handles random writes better than RAID5.

 

A little quote from here

Hitachi does not recommend using RAID 6 drives with high performance applications where extreme random writes are being performed. In some cases, the use of RAID 1 or RAID 1+0 is essential. 

From a performance standpoint, the performance of RAID 5 and RAID 6 is pretty similar when we talk about Random Read, Sequential Read and Sequential Write workloads. There is added penalty when we talk about Random Write workloads, that is because of the two dimensional parity. Compared to RAID 5, RAID 6 takes a 33% Performance hit on Hitachi with Random Write workloads. 

 

My opinion - go for RAID10 if you are using an onboard controller and able to get a fourth drive. Chances are you'll never ever ever experience an HDD fail during the rebuild of RAID5, but I'm kinda being paranoid here. RAID10 and 3-disk RAID5 have approximately the same amount of useful space, 4-disk RAID5 has a little more.

 

 

Thank you..  

[IdeaStormer]

Well if you truly want to be safe and are limited by the three drives you can only really do RAID 1, with the third drive as the spare. If you use RAID 5 and one dies you need to go out and buy a new drive that matches or is greater in size than what you have.

 

So in keeping with the original question/constraints, RAID 1.

[SMURG]

RAID 5 on >1TB drives is, mathematically speaking, a pretty bad idea. The probability of a second drive failure during the rebuild process becomes an issue once you hit 2TB per drive, so it can only be worse for 3TB drives (more data = longer rebuild time = higher chance of concurrent drive failure). Somebody somewhere worked out the probability based on how many errors drives are rated for per amount of data, and found that after 2TB RAID 5 becomes unfavourable. TLER helps if you're using something like WD Reds, so 2TB per drive may be alright in that instance, but 3TB? Idk, your mileage may vary

Personally, when it comes to RAID I've only ever used 500GB or 1TB drives partially for this reason. I might do a RAID 1 of some 6TB drives once they come out though, if it makes economic sense anyway.

[IdeaStormer]

RAID 5 on >1TB drives is, mathematically speaking, a pretty bad idea. The probability of a second drive failure during the rebuild process becomes an issue once you hit 2TB per drive, so it can only be worse for 3TB drives (more data = longer rebuild time = higher chance of concurrent drive failure). Somebody somewhere worked out the probability based on how many errors drives are rated for per amount of data, and found that after 2TB RAID 5 becomes unfavourable. TLER helps if you're using something like WD Reds, so 2TB per drive may be alright in that instance, but 3TB? Idk, your mileage may vary

Personally, when it comes to RAID I've only ever used 500GB or 1TB drives partially for this reason. I might do a RAID 1 of some 6TB drives once they come out though, if it makes economic sense anyway.

 

Here you go, 6TB Drives : http://www.seagate.com/www-content/product-content/enterprise-hdd-fam/enterprise-capacity-3-5-hdd/constellation-es-4/en-us/docs/enterprise-capacity-3-5-hdd-v4-ds1791-3-1403us.pdf

 

Let us know when its ready

[SMURG]

Here you go, 6TB Drives : http://www.seagate.com/www-content/product-content/enterprise-hdd-fam/enterprise-capacity-3-5-hdd/constellation-es-4/en-us/docs/enterprise-capacity-3-5-hdd-v4-ds1791-3-1403us.pdf

 

Let us know when its ready

Yeah, I saw Seagate was bringing out those and I know Hitachi have those He6 drives on the way, but there's no word on pricing anywhere so I have no idea if it would make any sense to do lol

EDIT: £227 (once u factor in tax) by the looks of it: http://www.newegg.com/Product/Product.aspx?Item=N82E16822178520 which is a really good price. If they stay that low I'll definitely be picking 2 up at some point.

[Zeke]

Yeah, I saw Seagate was bringing out those and I know Hitachi have those He6 drives on the way, but there's no word on pricing anywhere so I have no idea if it would make any sense to do lol

EDIT: £227 (once u factor in tax) by the looks of it: http://www.newegg.com/Product/Product.aspx?Item=N82E16822178520 which is a really good price. If they stay that low I'll definitely be picking 2 up at some point.

Seagate in a failsafe RAID? 

[Blackterror25]

WD greens have a tendency of 'dropping' out of a raid when they go into low power consumption mode. Also for a RAID you need to have drives of equal size or ells you'll lose 2TB of that 3TB drive.

[IdeaStormer]

Seagate in a failsafe RAID? 

 

Well BackBlaze is not the file server standard for quality, sure its dirt cheap but quality hardware, nope. That factors in on all the fails, work with file servers in the Petabyes and I'd say 80% are Seagate and no way do we get any failures in that range, in fact we are actually surprised how low our failure rates are and these servers are 24/7/365 and are written to and read and data removed by the Terabyes, age in in the 3 to 4 years of the disks.

 

I do agree with there findings on the 1.5TB Seagate drives though, they suck hard.

[Zeke]

I do agree with there findings on the 1.5TB Seagate drives though, they suck hard.

Yeah, but I hope newer high-capacity models are more reliable. 

[mrbulle]

WD greens have a tendency of 'dropping' out of a raid when they go into low power consumption mode. Also for a RAID you need to have drives of equal size or ells you'll lose 2TB of that 3TB drive.

 

i was only going to do raid with the three 2tb wd's.. but a friend ass often issues with one disk fails or dropping out and thats a green.. so for now i'll drop the raid i guess.. 

 

any disks recommended for raid?