Doubling Theoretical Performance: RAID-0

For those of you who are already familiar with RAID and how it works, go ahead and skip to the benchmarks; these next two pages are designed to serve as brief introductions to the two most common forms of RAID on the desktop: RAID-0 and RAID-1.

Otherwise known as striping, RAID-0 is the only performance-enhancing form of RAID that we'll be talking about in this article. The premise behind striping is simple. Data being written to a drive is split into "stripes", generally 16 - 256KB in size, with each stripe being written to a different drive in the array. For example, say we were dealing with a 2-drive RAID-0 array with a stripe size of 128KB and we wanted to write 256KB of data; drive 0 would get the first 128KB of data written to it, and drive 1 would get the remaining 128KB.

Writing to a single hard disk

Writing to a two-disk RAID-0 array

Here, you can see that the write performance of RAID-0 can be almost double that of a single drive, since twice as much data gets written at the same time. The higher write performance is obtained at the expense of some controller overhead, since the RAID controller has to handle splitting up data into stripes before sending it to the drives themselves - but with modern day microprocessors being as fast as they are, the overhead is usually thought of as negligible.

Reading works the exact same way, but in reverse. Say that we want to read that same 256KB of data back; we pull one stripe from drive 0 and the other stripe from drive 1. The read is now completed in half the time, theoretically doubling performance.

We are careful to use the word "theoretical" because the performance advantages of RAID-0 disappear quickly if we're not dealing in ideal situations like the ones we just described. If too large of a stripe size is used, then the performance advantages of RAID-0 can be lost, while too small of a stripe size could result in excess overhead, reducing the performance improvement of the striped array.

We have seen in the past that for most desktop applications, the largest stripe size that a desktop RAID controller will offer is usually the best choice for performance. With Intel's ICH5/6, that translates into a 128KB stripe size, which for our comparison is what we decided to go with. The other stripe size options didn't offer any better performance for our desktop test suite.

The main downside to RAID-0, other than cost, is reliability. The size of a RAID-0 array is the sum of all of its members; so, two 100GB drives in a RAID-0 array will give you one array with a 200GB total capacity. Unfortunately, if you lose any one of the drives in the array, all of your data is lost and isn't recoverable. Since two drives are working in tandem and are both necessary to hold your data, you effectively halve the mean time between failure by moving to a two-drive RAID-0 array.

Index Putting the Redundancy in RAID: RAID-1
Comments Locked


View All Comments

  • Arth1 - Thursday, July 1, 2004 - link

    The article contains several factual errors.
    RAID 1, for example, does have *read* speed benefits over a single drive, as you can read one block from one drive and the next block from the other drive at the same time.
    Also, what was the block size used, and what was the stripe size?
    Was the block size doubled when striping (as is normally recommended to keep the read size identical)?
    Since non-serial-ATA drives were part of the test, how come THEY were not tried in a RAID? That way we could have seen how much was the striping effect and how much was due to using two serial ATA ports.
    All in all a very useless article, I'm afraid
  • qquizz - Thursday, July 1, 2004 - link

    here, here, what about more ordinairy drives.
  • Kishkumen - Thursday, July 1, 2004 - link

    Regarding Intel Application Accelerator, I would like to know if that was installed or not as well. It seems to me that could potentially affect performance quite a bit. But perhaps it doesn't make a difference? Either way, I would like to know.
  • pieta - Thursday, July 1, 2004 - link

    It's funny to see metion of ATA and performance. If you really want disk performance, get some real SCSI drives. Without tag cmd queuing, RAID configurations aren't able to reach their full potential.

    It would be interesting see hadware sites measure SCSI performance. Sure, ATA has the price point, but with 15K SCSI spinners so cheap these days, the major cost is the investment in the HBA. With people dropping 500 bucks on a video card, why is it so inconvievable to think power users wouldn't want to run with the best I/O available?

    I was suprised not to see any Iometer benchmarks. IOPS and response times are king in determining disk performance. Iometer is still the best tool, as you can configure workers match typical workloads.

    Show me a review of the latest dual ported ultra320 hardware raid HBA stripped across four 15k spinners. Compare that with a 2 drive configuration and the SATA stuff. Show me IOPS, response times, and CPU utilization. That would be meaningful, as people could better justify the extra $2-300 cost going with a real I/O performer.
  • meccaboy858 - Thursday, July 1, 2004 - link

  • meccaboy858 - Thursday, July 1, 2004 - link

  • meccaboy858 - Thursday, July 1, 2004 - link

  • meccaboy858 - Thursday, July 1, 2004 - link

  • Nighteye2 - Thursday, July 1, 2004 - link

    Of course, RAID 0 makes little sense for raptors, which are already so fast that they hardly form a bottleneck.

    RAID 0 makes more sense for slower, cheaper HD's...try 2 WD 80GB 8MB cache harddisks, for example. Together they are cheaper than a raptor, but I expect performance will be very similar, if not faster.
  • Taracta - Thursday, July 1, 2004 - link

    I am tired of seeing these RAID 0 articles just throwing 2 disk together and getting results that are contrary to what is expected and not dig deeper into what's the problem. I am only posting my comment here because of my repect for this site. Drive technology and methodlogy has to play apart in discussion of RAID technology. The principle behind RAID 0 is sound. The throughput is a multiple of the number of drives in the array (You will not get 100% but close to it). Not getting this, it should be examined as to WHY? One of my suspicion is that incorrect setup of the array is the primary culprit. How is information written to/from the drive, the array and to individual drives in individual arrays. What is the cluster and sectors sizes. How is the information broken up by the controller to be written to the array. Take for example each drive in a array has a minimum data size of 64bits and you have array sizes of 2 rives 128bits, 3 drives 192bits and four drives 256bits. In initializing you array do you intialize for 64bits, 128bits, 192bits or 256bit? Does it matter? Say for example you initialize for 64bits, does the array controller writes 64bits to each drive or does it writes 64bits to the first drive and 0bits (null spaces and wasting and defeating the purpose of the extra drives) to the other drives because it is expecting the array size bits (eg 128bits for 2 drives)or does it split the 64bits between the drives and waste space and kill performance because each drive allocate a minimum of 64bits. I was waiting for someone to examine in detail what's happening. Xbitlabs came close (from looking at the charts)that they could almost taste it I am sure but still jump to incorrect reasoning.

    I know I am rambling but in short the premise of RAID arrays are sound so why is it not showing up in the results of the testing?

Log in

Don't have an account? Sign up now