Kaydol

Flood göndermek, insanların floodlarını okumak ve diğer insanlarla bağlantı kurmak için sosyal Floodlar ve Flood Yanıtları Motorumuza kaydolun.

Oturum aç

Flood göndermek, insanların floodlarını okumak ve diğer insanlarla bağlantı kurmak için sosyal Floodlar ve Flood Yanıtları Motorumuza giriş yapın.

Şifremi hatırlamıyorum

Şifreni mi unuttun? Lütfen e-mail adresinizi giriniz. Bir bağlantı alacaksınız ve e-posta yoluyla yeni bir şifre oluşturacaksınız.

3 ve kadim dostu 1 olan sj'yi rakamla giriniz. ( 31 )

Üzgünüz, Flood yazma yetkiniz yok, Flood girmek için giriş yapmalısınız.

Lütfen bu Floodun neden bildirilmesi gerektiğini düşündüğünüzü kısaca açıklayın.

Lütfen bu cevabın neden bildirilmesi gerektiğini kısaca açıklayın.

Please briefly explain why you feel this user should be reported.

Why can’t mirrored drives also work as striped

I’m planning to build a 2 bay nas and my only options are raid 0 or raid 1. I would like the speed of raid 0 but that risks redundancy. So that comes back to my original question. What is stopping mirrored drives from being used as if they were striped. So there is both performance and redundancy and the only requirement is an even number of drives.

Benzer Yazılar

Yorum eklemek için giriş yapmalısınız.

5 Yorumları

  1. If it helps, Linux `md` (including LVM RAID as it basically uses `md` under the hood) will provide a performance boost on RAID 1 if performing multiple queued read operations. Write is still just as fast as one disk.

  2. > What is stopping mirrored drives from being used as if they were striped.

    ugly software designed in the middle of 90x, when the one drive was able fully saturate both bus and cpu. Yes, I mean linux md raid.

    Not only it had ability to gave you twice performance loss on write (which is also ridiculous because all modern hardware works asynchronously), but even on read it had no distributed reads, always sequentially doing it from single spindle.

    There were some attempts to make it more robust (so todays md not so extremely bad) but it still works badly (because actually it needs to be completely rewritten from scratch).

    zfs (sun one, not the modern nonworking opencrap) and freebsd geom mirrors both ARE using raid1 properly, distributing linear reads in parallel and multiple random (as it happens on multi-user multitasking system) by load. (geom also able to use other algos, for example, properly handle ssd+hd mirrors, always reading fastest one)

    Note also what most if not all modern “hardware” raid cards have linux under the hood and resolve to same md code 25 years old (worst part of it – because hardware vendors never updated they code from the one they stole 25 years ago)

  3. raid 50 and 60 are mirrored, striped arrays.

    raid 50 requires 6 drives, raid 60 requires 8 drives.

    striping splits a file, equally, between 2 drives (in your 2 drive example). adding a 3rd drive to a striped array allows you to go to raid 5 (striping with parity – keeps your data going even if a drive dies)). raid 6 is 2 parity drives (keeps your data going even if you lose 2 drives).

    since you state money is your issue, i would use a mirrored array for the redundancy for now, until you can afford more drives.

  4. Mirrored RAID depends on the specific RAID configuration and how it’s handled. RAID 1 *can* improve read performance. Some are better with multi threads (reading one file from one disk, while reading another file from the other), some can stripe and read files like RAID 0. But write performance is no better than a single drive, if anything it’s slightly slower because it has to make sure both drives are synced.

    If you really want benefits of RAID 0 with RAID 1 protection, go with RAID 10. That will be two sets of RAID 1 mirrored drives striped together in RAID 0.

    RAID 0
    |
    ———————-
    RAID 1 RAID 1
    | |
    ——- ——-
    | | | |
    DISK1 DISK2 DISK3 DISK4

  5. That’s what raid1+0 or raid10 does but you’d need 4 drives I think.