draid vs other implementations #11620
Replies: 1 comment 4 replies
-
|
IMHO draid main selling point is very fast rebuild times, when all disks
are busy rebuilding instead of conventional raid when rebuild speed is
limited by the write performance of the spare. Also all disks including
spares are in use during normal operation, when in the conventional raid -
spares are completely idle.
…On Sat, Feb 20, 2021 at 5:58 AM Alexander K. ***@***.***> wrote:
Hello everyone,
i use zfs for storing my data safe at work and at home. In the newest zfs
release you have announced the new draid feature. What i want to know is,
because several years before i came to zfs i used much more systems from
Netapp (FAS) and they use a Raid-4 implementation. I know that this type of
Raid would be very slow if writes are not cached to a nvram device and the
parity disk would limit the write througput. But you get another nice
features like expansion of the raid group, because all parity is stored on
one or like the curent implementation three drives. With a write cache ssd
this would be very nice to have. Or do I misunderstand how it works here?
But what makes draid better than raid-4, dp or tec?
Thanks in advance
NR
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#11620>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABXQ6HPVPI2QUYKXI34GMA3S72YHFANCNFSM4X43XUWA>
.
|
Beta Was this translation helpful? Give feedback.
-
|
Hello, thank you very much, now I understand a little bit more about this. That is probably one of the reasons why Netapp introduced "Physical Mirror (plexes)" from raid groups at the time, so that the data is not lost during the rebuild if more than 2 or 3 hard disks break. But a feature to mirror the entire raid group would also good thing for zfs, because then you can safely use raidz1 or raidz2 without losing the whole pool if more than two drives fail. It would be right that you can mirror the data asynchronously, but if the pool fail all applications that are run on it would also fail. That could prevent this. |
Beta Was this translation helpful? Give feedback.
-
|
NAME SIZE ALLOC FREE CKPOINT EXPANDSZ FRAG CAP DEDUP HEALTH ALTROOT |
Beta Was this translation helpful? Give feedback.
-
|
pool: tank errors: No known data errors |
Beta Was this translation helpful? Give feedback.
-
|
@jammin2night There is no distributed spares in the pool, so the rebuild time is limited by the time to copy all the data to the replacement drive, which is IMHO the main selling point of draid. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Hello everyone,
i use zfs for storing my data safe at work and at home. In the newest zfs release you have announced the new draid feature. What i want to know is, because several years before i came to zfs i used much more systems from Netapp (FAS) and they use a Raid-4 implementation. I know that this type of Raid would be very slow if writes are not cached to a nvram device and the parity disk would limit the write througput. But you get another nice features like expansion of the raid group, because all parity is stored on one or like the current implementation three drives. With a write cache ssd this would be very nice to have. Or do I misunderstand how it works here? But what makes draid better than raid-4, dp or tec?
Thanks in advance
NR
Beta Was this translation helpful? Give feedback.
All reactions