Post your hard drive benchmarks!

write back DISabled

2 x 500Gig 32meg cache 7200.11 in raid 0 @ 64kB

write-backDISabledshort.jpg
 
Last edited:
Mint_Sauce said:
It's 4x500gb as the storage drives, the fifth 500gb just does the parity stuff so if one of the four drives fails I can just replace it with another 500gb drive and it will carry on like nothing happened. That's why it's worth it.
Click to expand...
Just to be pedantic what you've described is RAID3 (dedicated parity disk), with RAID5 the parity is striped across all drives so in a 5 disk array each drive is 80% data and 20% parity. The advantage of striping the parity is that the RAID5 array is not limited by the write speed of a single drive in the way that RAID3 is. However to take advantage of this you need a controller capable of calculating the parity bits at a sufficiently high rate to keep up with the disks, that usually means splashing a lot of cash.
 
Before I get any 'that's really slow', The pic below is of my main array on my home server. As you can see smashes through the 2048gb barrier and as I have kept it as one physical drive/partition these fairly heavy speed drops were to be expected:

array.jpg


As it is my home server and hosts all my movies/everything else under the sun I wasn't bothered about flat out speed, merely the huge single drive as I use a LOT of space. The array is well over half full already. 11xWD5000KS drives in raid 5 with a hot spare. To be honest the speeds shown on that test don't do it justice IMO as stuff really flies up and down the gigabit between the machines and we have never had the slightest hint of a bottleneck :).
 
Plain PCI? That would be a hanging offence in my book! :D. The board is an Asus P5WDG2 WS PRO with PCI-X 133 slots :). Usual thing of all hardware that isn't used is disabled as in USB etc. Onboard raid is being used for the OS drive and the the only other card in there is an x300 pci-e graphics card.

http://www.microsoft.com/whdc/device/storage/LUN_SP1.mspx - 32 bit Server 2003 R2

It is running a GPT disk on . While I was never impressed with the 'on paper' speed, it runs perfectly well with 100% reliability and is a stupid amount of storage. The only way to see for sure would be to destroy the array and either create 2 much smaller arrays, OR enable auto carving, but I don't have a spare few TB's about the place to move the data to ;).
 
Last edited:
Seagate Barracuda 7200.10 160GB ATA-100 8MB Cache OEM IDE 7200rpm
(No RAID or anything fancy). Brand new, one of a pair.

ST3160815A.jpg



Finally spent time figuring out how to post images. Been meaning to do that for ages. :)
 
Last edited:
rpstewart said:
Just to be pedantic what you've described is RAID3 (dedicated parity disk), with RAID5 the parity is striped across all drives so in a 5 disk array each drive is 80% data and 20% parity. The advantage of striping the parity is that the RAID5 array is not limited by the write speed of a single drive in the way that RAID3 is. However to take advantage of this you need a controller capable of calculating the parity bits at a sufficiently high rate to keep up with the disks, that usually means splashing a lot of cash.
Click to expand...

Thanks for explaining that. I can see why it takes so long to migrate now. All the data needs to be rebuilt over all the drives and the parity built. I just thought the 5th drive was plonked in and the parity was then put on that drive.

I'll post up a benchmark when it's finished for a direct comparison between a 4 disk array raid 0 and a 5 disk array raid 5.
 
You must log in or register to reply here.
Back
Top Bottom