• 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.

RAID 1 performance

M

MassiveOverkill

[H]ard|Gawd
Joined
Jul 8, 2005
Messages
1,748
Aside from the extra boot time required to get he RAID array going, If I run two 160 Seagate 8 meg cache SATAII w/NCQ drivesin RAID 1, will I see a decent performance increase in reads over a single drive? Will it be similar in performance in reads over the same config in a striped array?
 
afaik raid 1 doesn't have any performance gain, nor penalty. all the controller is doing on writes is sending the same command to two drives. On reads, it reads both and compares. AFAIK anyway. mage might know reads better.
 
From what I understand, Parity on a RAID drive is not used to verify data integrity, but only for reconstructions efforts. This means that RAID-1 will only read the data from one drive and there is no comparison during normal operation.

Some "smart" controllers would be able to increase the STR of a RAID-1 array by reading from it in a manner similar to RAID-0, however I do not know which controllers do this. The idea would be to read even block from one drive and odd ones from the other. Since both drives should contain the same data, this may effectively double the STR, however I am not sure if there isn't a mechanical limit to this, since skipping blocks may be associated with a rotational latency penalty.

RAID-1 could also facilitate decrease in read-seek times, if the controller was set up to do this. By placing the two independent RW-heads at the 1/4 and 3/4 positions, the average seek distance would be decreased to 1/4 of the drive. This assumes that the seeks are completely random, which means that for a single disk the expected seek distance would be 1/2 the maximum distance.
Once again, I do not know if any controller does this. I believe that doing such a thing may also increase the average seek time for a write, since it is most likely that one head is further away from the writing prosition than the other. If the controller needs to wait for both heads to be able to write a specific block before issuing the command, the write seek time would increase.

Disclaimer: I do not know how any controller implements its RAID-1 and I don't know that much about mechanical limitations that would prevent either of these optimizations to work. My ideas are based what could be done 'theoretically' and may prove to decrease performance in a real-world scenario.
 
GLSauron said:
afaik raid 1 doesn't have any performance gain, nor penalty. all the controller is doing on writes is sending the same command to two drives. On reads, it reads both and compares. AFAIK anyway. mage might know reads better.
Click to expand...



From what i had read , READ performance is increase on a raid 1 as it can read data from either or both drives, but of course writes are slowed down for obvious reasons.

thanks for the clarification.
 
GLSauron said:
afaik raid 1 doesn't have any performance gain, nor penalty. all the controller is doing on writes is sending the same command to two drives. On reads, it reads both and compares. AFAIK anyway. mage might know reads better.
Click to expand...
I don't know of any controller that reads from both disks and compares. Modern disks either read the correct data or return an error, there's a very, very low chance of returning incorrect data and asserting it's right. This is due to ECC and Hamming codes on the hardware level.

Depending on implementation, the controller could read from both disks, or just one. The former is obviously the better option, but it's a bit more difficult to write the software for.

 
So, I would probably have to use an add-in controller card to get the benefits as onboard RAID built into today's motherboards won't do smart RAID and read from both disks simultaneously to increase performance?
 
MassiveOverkill said:
So, I would probably have to use an add-in controller card to get the benefits as onboard RAID built into today's motherboards won't do smart RAID and read from both disks simultaneously to increase performance?
Click to expand...
There is no guarantee that an add-in controller would do such a thing. If you are really, really interested in this, you're going to have to research it yourself, i.e. write the manufacturer of the add-in or onboard solution that you are considering and hope that you won't get turned away with some marketing BS. Maybe you will get lucky and some of the people on here with industry experience that tell you which controllers support the feature that you are looking for.

On a sidenote: even if the controller is extremely smart about the way in uses multiple disks to its advantage, it is not guaranteed that the employed strategy will benefit you, since it is most likely application dependent.
 
RAID-1 increases performance for small reads at random locations, but only if you have decent RAID-1 software. From what I have seen onboard SATA raid doesn't do clever raid-1.
 
You must log in or register to reply here.
Back
Top