2.5" HDD Raid

OK, running out of space in my MATX case, want a bit more speed on my system drive.
Raptors look nice but expensive.

Question is - does anyone know of the performance of a couple of 7200-10000 rpm 2.5" drives in raid 0?

Any ideas on pros/cons of this type of setup would be great.

10 answers Last reply
More about raid
  1. If you simply Google RAID 0 performance, you will get over 2000 links of reviews and charts.
  2. A Velociraptor is 2,5" but comes with a 3,5" bay to prevent overheating. If you remove the heatsink and use it as a normal 2,5" drive, it will fail and your warranty will be void.

    RAID0 performance means double STR and increased IOps, when properly setup. Know that performance varies per RAID implementation - no RAID driver is the same. Intel ICHxR southbridge allows you to use RAM buffercache (advertized as 'write cache') which makes it faster than other implementations, but also more dangerous in the case of unclean shutdown.
  3. Thanks

    OK - same question really, assume I know about RAID performance (didn't know about google though - thankyou for the tip!) and cannot fit raptors into the case/do not want to void warranty etc.

    Anyone tried this? The only specific review I can find is from toms dated 2002.

  4. You mean RAID with normal 2,5" drives? If you know about RAID0 you know you can expect 2*STR speed and increased IOps when properly configured.

    If you want ultimate speed then you can always buy two good SSD's and put them in RAID0. SSDs are often 2,5". But then, it depends if you can justify the cost with the limited capacity available. Yet, such a setup should be quite safe even with RAID0.
  5. enlightenment, thanks for your input. SSD's would be nice, but the price is just unjustifiable for me right now. I was thinking of using 3 or 4 drives for the extra speed, so the transfer rate should scale a bit beyond 2* - I was hoping someone on the forums had tried this and could give an idea of the heat / noise 4x2.5" hdd's in raid 0 generates, may have to just do it and see what happens.

  6. Many 2,5" drives are quite silent, because they are targetted towards use in laptops. Just lookup the specs on the product pages, 25dBA is quite silent. Some drives you can hear the seeks others you barrely hear them, that also affects performance since seeks can be faster if more power is used but also generates more noise. Some disks feature AAM, which allows you to prioritize between lower seek performance and less noise - or maximum seek performance with more noise generated and some more power consumption.

    With 4 disks in a proper RAID0 setup you achieve 4*STR speed. But sequential speed above 100MB/s are nog always useful, since you'd need another volume with at least the same speeds to achieve this speed in reality. And gigabit won't help you either, since it about performs like a single hdd.

    To get max RAID performance:
    - pick a controller with PCIe or embedded bus, such as true onboard chipset sata ports or addon PCI-express controller, never use PCI or USB (which also connects to PCI bus).
    - craft your partitions to fix any stripe misalignment
    - set the stripe size high enough to allow one I/O to be handled by one drive
    - enable write caching (buffercaching) if using ICHxR RAID drivers
    - pick optimum filesystem settings when creating the FS on the RAID volume
    - benchmark the volume using proper tools (i.e. not HDTune but something like ATTO-256)

    I do have to warn you that 4 notebook drives in RAID0 presents a very viable risk to data loss. Make sure that, in any point of time, one disk can fail without causing you alot of headaches. When using RAID0 that either means the data it stores is not really important (i.e. can be re-downloaded or re-generated) or that you have a proper backup in place.
  7. There is generally no real world(vs. synthetic transfer rate benchmarks) performance advantage to raid of any kind.
    Go to www.storagereview.com at this link: http://faq.storagereview.com/tiki-index.php?page=SingleDriveVsRaid0
    There are some specific applications that will benefit, but
    gaming is not one of them. Even if you have an application which reads one input file sequentially, and writes
    it out, you will perform about as well by putting the input on one drive, and the output on the other.

    For the best system drive, look at the velociraptor 150gb or 300gb. It does make a difference.
    Go to www.storagereview.com and access their performance database. You will see than the velociraptor comes out on top(excepting the mtron SSD) in the office drivemark 2006 benchmarks. It is the most representative of everyday usage as compared to synthetic benchmarks which only measure transfer rates in an optimum environment.

    If 300gb is not enough, just add a high capacity storage drive.
  8. StorageReview uses a PCI controller with a specific low-end driver RAID implementation to prove RAID0 does not provide any benefit? There's a lot of blabla in there without hard data to back it up. The little data they gathered proves even in the worst case scenario, PCI limited driver-RAID with stripe misalignment, some applications provide performance increase while none show a performance decrease. If SR has to prove anything, let them retry with a proper RAID setup first - this is ridiculous for a site known as "storagereview". IMHO ofcourse, but i do have some experience with testing RAIDs myself.

    The article to mention was fuel for tweakers.net to write a counter-article proving that under the right circumstances raid0 can provide real benefits to desktop users. The conclusion of this article:

    AnandTech and Storage Review should be wise to investigate matters more thoroughy before jumping to quick conclusions. You don't judge a Porsche on its capabilities to carry groceries. A car like that serves a different purpose, and it should be judged on that instead - even if it will never fulfill its true purpose in real life. Power users, tweakers and hardware enthusiasts, the target audience AnandTech, Storage Review and Tweakers.net try to please, use their desktop systems in a different way than the pretty blonde next door who only uses it to check her Hotmail account. What we're trying to say is that you shouldn't assess the performance of RAID 0 with benchmarks that are not made to test the performance of the storage subset. AnandTech and Storage Review's negative verdict on RAID 0 in the desktop environment will likely have a profound influence on the opinions of uninformed users for years to come. A sure loss, since their verdict couldn't stand up to trial.

    The original article is Dutch, but an English version is also available here:
  9. No doubt, that a high end raid card will help performance. Most of the users posting here are asking about the benefits of using their onboard raid controllers.
  10. The onboard SATA-controller doesn't communicate via PCI but has a direct "embedded" interface to the southbridge. That means maximum performance, no cap.

    The benchmarks Anand & SR provided were using an PCI addon fake RAID controller, thus being a flawed setup no one will actually use. But then i guess, maybe they just wanted the results to be low.

    Whether you pick driver RAID or Software RAID, as long as you don't hit any interface bottlenecks (PCI) you should be fine. Some motherboards have additional onboard controllers not supplied by the chipset, like Silicon Image, Adaptec or JMicron. Those can be both connected to the southbridge via PCI or PCI-express. You can sometimes recognise them as their SATA ports often have another colour.

    My old Asus K8N4-E Deluxe motherboard used in my original NAS system is one of these boards which provide 4X SATA on the chipset controller and an additional 4x SATA via Silicon Image 3114 PCI SATA controller. The latter being much slower, something i can test very easily. So always avoid PCI and use either your onboard SATA controller, or an add-on controller connected via PCI-express.
Ask a new question

Read More

NAS / RAID Hard Drives Storage Product