RAID 1 speed benefits

[Fike]

I have been spending some time trying to tweak my machine to get a bit more performance out of Photoshop CS3. I came to a few conclusions. 1) having the scratch disk and pagefile on a different partition of the boot disk has minimal benefits. 2) RAID 1 read speed is quite fast while write speed is pretty slow.

So, what am I doing about it?

I am getting more RAM to avoid the pagefile as much as possible. For the times when that is impossible, I am getting another hard drive that will be on a dedicated channel and used only for scratch disk and pagefile.

So about that RAID 1 speed thing. I have my RAID configured for redundancy. This should offer no speed advantage, but my guess is that during read cycles it can actually read off both hard drives simultaneously in a striping-like operation, it doesn't need to actually read both hard drives, it only needs to read the alternate long enough to verify the contents are still valid. Put another way, the RADI card could use 1st cycle to read segment 1 off hard drive 1 and segment 2 off hard drive 2. During the second cycle, it can read segment 3 off hard drive two and segment 4 off hard drive 1. As long as the two drives addressing stays in synch, the RAID will know that everyting is okay on both drives. The reason that I am making this hypothesis is that during read operations (measured by Sandra 2005) my setup acheives about 64 Mb/s while during write operations it reaches a more modest 30 Mb/s. These times are achieved on cheap ATA 100 hard drives, so the 30 MB/s is expected while the 64MB/s is pretty darn good.

I thought it was interesting.

 

[MobiusPizza]

You are right in that RAID 1 has nearly double reading speed compared to writing speed. While I noticed this before, I do not know the details as of why. But your assumptions seems logical

 

[SuperNaruto]

better raid 1 implementation can read data from both drive since they're suppose to be the same... thats why r1 is better at read.. not if it's implemented

http://www.acsdata.com/raid-1-data-recovery.htm

When reading, both disks can be accessed independently. Like RAID 0 the average seek time is reduced by half when randomly reading but because each disk has the exact same data the requested sectors can always be split evenly between the disks and the seek time remains low. The transfer rate would also be doubled. For three disks the seek time would be a third and the transfer rate would be tripled. The only limit is how many disks can be connected to the controller and its maximum transfer speed. Many older IDE RAID 1 cards read from one disk in the pair, so their read performance is that of a single disk. Some older RAID 1 implementations would also read both disks simultaneously and compare the data to catch errors. The error detection and correction on modern disks makes this less useful in environments requiring normal commercial availability. When writing, the array performs like a single disk as all mirrors must be written with the data.

 

[Matthias99]

RAID1 in general does not verify the data in real time while reading. This may be an option on some controllers. 60MBps is about right for a modern 7200RPM IDE drive when reading/writing sequentially. I'm surprised your performance is so low when doing sequential writes, though. What kind of controller is it?

For sequential reads, what you describe wouldn't really help anyway, since the drives would be constantly seeking; it's actually more efficient to just read sequentially from one drive.

You may get better performance during highly queued random reads from RAID1 (say, if you use it as the drive hosting a website). Most newer controllers will interleave I/Os to the drives, so you can read different things from each drive in the RAID1 array simultaneously.