• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Does software RAID really suck this much?

L

llars

Weaksauce
Joined
Jan 12, 2006
Messages
104
I just bought four 300gb SATA drives and I am runing them in RAID-5. The drives are Seagate 7200.8's The controller is the Silicon Image chipset on the A8N-SLI. I also have 2 Raptors in Raid-0 on the Nvidia Chipset. here are the benchmarks I got.

Raptors:
Sequential Read: 51.8 MBytes/Sec
Sequential Write 121.4 MBytes/Sec
Random Seek + RW 6.4 MBytes/Sec

RAID-5:
Sequential Read: 21.8 MBytes/Sec
Sequential Write 5.5 MBytes/Sec
Random Seek + RW 2.70 MBytes/Sec

IDE Drive:
Sequential Read: 17.3 MBytes/Sec
Sequential Write 27.3 MBytes/Sec
Random Seek + RW 2.78 MBytes/Sec

Do those numbers seem right, or is something else going on?
 
Well the Raptors are hella fast drives, so nothing really compares to it. But your write speeds are very low.
 
All your numbers except sequential write for the Raptors look terrible. I'd be inclined to blame this on something Else. When did you last reinstall Windows? What drivers are you using? Run spyware check lately? Antivirus set to UberParanoid?

 
Windows install is 5 or 6 months old. But come to think of it, awhile back I replaced a busted Deluxe board with a Premium. I figured since all of the chipsets are the same, I would be fine. I have not had any problems. The only issue I have is writing to the RAID-5 array.

I think I'll blow everything out and see what happens...
 
I had 4 x 400GB In software raid5 on xp pro.

My sequential writes were ~ 7MB/second but my sequential reads were about 120-130MB/second (pci-x, not pci-e) non-raid promise controller.
 
All your read speeds should be higher than that, significantly, even on the single drive. I've never had a single drive write better than it reads. Read speeds on the raid 5 array with 4 drives should be close to a 3 drive raid 0 array, but slightly slower. Your raptor read speeds is less than the read speed of a single raptor. They usually push closer to 60-70 MB/s not in raid.

Writes shouldn't even be *that* low on the raid 5.

You may definately want to track down new drivers and consider doing a lot of system analysis to find out what's going wrong.
 
Writes on a RAID 5 array are always much slower than reads. This is true in the case of both hardware and software RAID 5 arrays (though much more pronounced in a software array).

Let's say you're doing a write to a RAID 5 array with 3 drives. To perform the write that is not exactly the size of the stripe (most writes will be like this). The controller has to read the data stripe from two of the drives and the parity stripe from the third drive. It then has to modify the part of the data it needs to, then perform an XOR operation on the new data to get generate the parity stripe. Then it has to write the change portion of data and parity stripe back to the drives.

You can see how this is much more intensive than a non-parity-protected write. In the case of software RAID, not only does all this have ot happen, but the data has to go both directions across the system bus into memory, then have the host CPU perform the XOR, then write it back to the controller which writes it back to the drive. Software RAID 5 generates a lot of exter traffic on the PCI bus, memory bus, and front-side bus to the CPU.
 
I reformatted and my write speed on the RAID-5 is at 6.5. I understand that RAID-5 will be slower, but I am wondering if it should be this slow.
 
All generalizations are false!

I guess windows software raid just does suck that much.

Wait, what benchmark are you using to generate those numbers? Try HDTach and post the results. Burst speed, sustained, latency are the important values.

 
I don't remember the name of the program offhand, I found it on google. I am not using WIndows based RAID. I configured the array in the bios of the SI controller, but I have heard from mutiple sources that the NV RAID is hardware based and the SI RAID is software based. On a side note, when I am experiencing this sluggish performance, CPU utilization is next to nothing. But accessing data on other disks is sluggish too when I am writing to the RAID-5 array.

I will download that util tonight and try it out.

Other specs if it makes a difference:

Asus A8n-SLI Premium
AMD X2 3800+
2GB Corsiar ValueSelect
 
Well I have about had it with this SI RAID...

I installed a SCSI card for a Tape Drive tonight and without knowing it I unplugged one of the SATA cables. Didn't see it because of the Video Card. Computer boots up fine, no warnings at all. (??????) A few reboots later I notice the RAID's bios is only showing three drives instead of four. So I plug in the fourth drive and Power up the system. RAID bios shows all four drives but only three in the Array. So I go into the Bios. And I find out that THERE IS NO #%$&^*& OPTION TO REPAIR THE ARRAY!!!

So now I am in the process of moving all of the files off of the array. I can't even use the fourth drive currently because it sees it as being part of a failed Array but it will not let me do anything with it. I'm just going to run the four drives independant of each other.
 
Final update. I am now running the drives in JBOD and the Sequential Write time is 44.6 Mbytes/Sec. So I guess software RAID really does suckthat much...
 
You must log in or register to reply here.
Back
Top