GA-X48T-DQ6 Slow Raid5 setup

[MasterX]

Hi,

Now a proud owner of the GA-X48T-DQ6 mobo I recently setup 2 raids on my system. Both on the ICH9R chipset.
2 samsung disks in raid 0 which work better than I ever expected and 3 seagate 500G disks in RAID 5, these not working so well.
I knew that RAID5 was going to be slower but I only get an average write speed of 20MB/s. Is this normal or could one of the disks be having problems and in this case be slowing them all down?

If there's a problem what would be the best method in finding this out? Most tests I did involved reading speeds, and that's working alright.

[Lsdmeasap]

I am not so sure about RAID5 Speeds but that does sound slow to me.

What are you testing with? Can you test with this and report back >>>
HdTach 3.0

I do read it is slower on Writes and here is why >>>

RAID 5 comprises a logical volume, based on three or more disk drives, that generates data redundancy to avoid the loss of an entire volume in the event of disk failure. The RAID controller creates a special parity block for each stripe of information, as Figure A shows. (Features at the OS level, such as Windows 2000's—Win2K's—disk striping with parity, can also perform this function.) The parity is typically a binary exclusive OR (XOR) operation—indicated here with the ~ symbol—on all the data blocks of the stripe. In Figure A, the RAID controller calculates parity as S3 = A3 ~B3 ~D3 ~E3.

When a write operation occurs on a RAID 5 volume (e.g., on block B3), the controller must update parity block S3. Because the controller must read all the blocks in the stripe to recreate the parity block, most RAID 5 controllers will go through the following steps, in which single quotation marks and double quotation marks represent modifications:

1. Read the block that needs modification (B3).
2. Read the parity block (S3).
3. Remove the knowledge of block B3 from parity S3 (S3'=S3~B3).
4. Calculate the new parity (S3"=S3'~B3').
5. Update the data block (B3').
6. Update the parity block (S3").

In other words, one application I/O requires four disk I/Os, and these four I/Os occur on two spindles, potentially disturbing other transfer operations on those volumes. . . .

Short random writes to a RAID-5 are VERY slow, but long sequential writes to a RAID-5 “should” be pretty fast, even approaching the performance of RAID-0.

But I did not find anyone stating better speeds than you without a Hardware Based Controller. Without a XOR Processor RAID 5 will not be very fast at all at writes

[Vertigon]

I had a similar problem recently. Things felt alot slower when I striped 128K on RAID0 over 2 drives using the Intel chip. My suggestion is to move one volume to the gigabyte controller, this might work the IRQ better. For those using Vista try HD tune:
http://www.hdtune.com/

Hello Lsdmeasap

[Lsdmeasap]

HD Tach RW works in vista, sorry forgot to mention you should run in XP Compatibility mode.

Ya, GA chip is faster but he needs RAID5 which it wont do. But like you mentioned stripe size IS also going to matter

Hey Vertigon, Zup

[svcglobal]

Originally Posted by Vertigon

I had a similar problem recently. Things felt alot slower when I striped 128K on RAID0 over 2 drives using the Intel chip. My suggestion is to move one volume to the gigabyte controller, this might work the IRQ better. For those using Vista try HD tune:
HD Tune

Hello Lsdmeasap

Trying not to hijack the thread, I found EP35C-DS3R ICH9R @RAID0 faster than Gigabyte controller (purple connector).
Here is the Intel result:

[Lsdmeasap]

Vs. Gigabyte results with same stripe size image please!

You can save results and make it show 2 comparing each other in different colors like shown here >>>



But yeah, Like I always say everyones results wont always be the same as mine as hardware always differs.

How is your burst rate 1000's times faster then SCSI?

[MasterX]

Thanks for the replies!

The problem with these benchmarks is that they only test read speeds, as I said reading from the raid5 is fine. (not great but an average 85MB/s)

The pro version of hd tune has a write test, but the disk needs to be empty.
Might try this later on but I'm not sure what I would learn out of this, because I did some other tests already.

I benched all disks before I put them together in raid, all of them had about the same min/max/avg, so I recon it's not a disk malfunctioning.
I did a manual write test by copying a HD movie to the raid5 with teracopy(and vista copy) which told me an average speed of 20MB/s.
I can only compare this to a friend of mine, he has a ASUS Striker II Extreme mobo. And he can achieve a write speed of 45MB/s, not great as well but I could live with it.
The mobo is more expensive, so I suppose it's the difference in chipsets? I would imagine intel being better at raid configs than nvidia..

[Lsdmeasap]

Ya the intel would be better then the nvidia chip. Sorry I guess I overlooked you meant write not read

What stripe size are you using? Smaller will be faster likely. If you are using 128 try 64 or 32

[svcglobal]

Originally Posted by Lsdmeasap

Vs. Gigabyte results with same stripe size image please! Could I connect these two Seagate HD to the Gigabyte directly? No need to reformat everything? No loss of data? If yes, I will do it right now.

You can save results and make it show 2 comparing each other in different colors like shown here >>>



But yeah, Like I always say everyones results wont always be the same as mine as hardware always differs.

How is your burst rate 1000's times faster then SCSI?

I don't know. I only did the Vista F6 installation with the last Intel drivers. Since I did the Vista setup many times I realized that Gigabyte controller took the double of the time.

[Lsdmeasap]

No you have to either Mirror your system and reinstall RAID array again on the other controller, or you have to just save the file for comparision and still reinstall on the other controller.

No, you can't just switch them between the two for sure!

What Stripe or Block size are you using for your RAID?