MSI P6N Platinum - Suboptimal RAID 0 performance (RAID 1 is faster??)

[Gohoos81]

After installing Vista Ultimate 32 bit on a RAID 0 (16 KB stripe size, 2x 320 GB Seagate 7200.10 SATA II), the transfer rates seemed to be poor. Windows Experience Index only gave the array (bootable array) 5.4. My other RAID configuration on the same desktop, a RAID 1 of 2x 250 GB WD 7200 rpm, was faster.

I downloaded and ran Disk Benchmark 2006 and benchmarked the disks, confirming my suspicion. The RAID 1 array is the "performance" setup on my system, and the RAID 0 lags behind ... considerably. I have attached the benchmarks as a pdf and included detailed system specs on the pdf sheet.

http://www.geocities.com/e6600g80gts/diskbenchmarks.pdf

SATA II compliant cables were used for all connections. The jumpers on the back of the Seagate drives for RAID 0 were previously removed to enable SATA II operation. All SATA and power connections were checked and verified.

 

[Gohoos81]

1. I installed the new nForce (15.01) drivers in Vista Ultimate 32-bit last night, then rebooted.

2. I benchmarked the WD 250 GB 7200 rpm RAID 1 array and using HD Tune and found a maximum sustained transfer rate of 65 MB/s, and an average seek time of 13 ms.

3. I then used the NVIDIA MEDIASHIELD/STORAGE utility and migrated the two WD 250 GB 7200 rpm HDD from RAID 1 (non bootable) to RAID 0. I then rebooted.

4. I ran HD Tune benchmarks on the NEW RAID 0 WD250 GB 7200 rpm disks and found a 32 MB/s maximum sustained transfer rate, about HALF of the transfer rate while the disks were in RAID 1.

--I cannot figure out what is going on.

ALSO:

5. the NVIDIA SMART disk monitoring utility showed for two hours last night, that it was "predicting a failure" on of the SEAGATE 320GB 7200.10 drives in RAID 0, which is the bootable array. After a period of two hours, the NVIDIA SMART disk monitoring utility showed the health status of the "failing disk" to be OK (same heatlh status as my three other HDD).

6. I REBOOTED, and then checked the NVIDIA MEDIAHSIELD SMART disk monitoring utility, and it showed the health of the disk to be "OK", although an hour earlier it has "predicted a failure" on that disk (disk 1.1, or connected to the 2nd SATA port on the mobo).

I CANNOT FIGURE THIS ONE OUT EITHER! Help would be appreciated for numbers 1-4 (question 1) and numbers 5-6 (question 2). Thanks, Cooper Wriston

 

[choirbass]

well... i would just make a recommendation to go without raid entirely (unless you can make specific direct use of the higher STRs, a lot of media editing for example, and to even then only host data on it that you wouldnt mind losing)

most windows situations however do not utilize the higher STRs at all well, most dont tbh, where 1 hdd will be about as fast as 4 or more hdds in raid 0, for most practical situations), and to instead just use the 2 hdds seperately for redundancy of important data... you may end up saving yourself a lot of grief in the long run too... incase one of the hdds does happen to fail, as it sounds like (unusually and unnacceptably slow transfer rates at times, failure alert one minute, and then not showing up the next)

edit: as far as the differences in STR performance between the different arrays... the problem might notve shown up when using raid 1, because youre not dependant on both hdds operating in sync for it to work properly (one of them could fail, and you might not even know it, and you can continue using your system without interruption... though the problem with not knowing something happened, is that you may still think you have 2 working hdds, but you might then only have 1 in that case, so no redundancy, so either way you would want to swap the failed hdd then to rebuild the array as soon as you find out about the failure, this is more ideal for servers though, so you can minimize downtime, but should never be a replacement for regular backups)... for raid 0 however, both hdds are required to operate in sync... and if theyre not, youll have varying problems... which may be what youre seeing, which is why failure seems possible by the symptoms you gave too.

 

[russki]

Whatever choirbass said. As in what made you think you would get better performance (other than trasfer rate) in the first place?

 

[Mondoman]

Yes, I think cb's suggestion of a drive on the verge of failure makes sense. r, I think the OP's goal in going to RAID 0 was in fact a higher transfer rate.