RAID's and New File systems

[raksasas]

RAID6, RAID 10, ZFS, BTRFS ???
I have been going over this thread and this article/comments. Between them,they mention "RAID's (6/10)" and "new file systems(ZFS/BTRFS)". I'll just say I have a headache coming on trying to get a grasp on this.

The reason why I am trying to get an upstanding of this is I am looking at replacing my RAID5 system that I built a couple of years ago. After that is I am looking at my backup solutions which I hope to better from the information that I gather from this.

 

[idea]

I don't understand why a hardforum member in his 8th year is asking to be educated and trying to get an "upstanding" on a topic that is easily googleable and well documented.

ZFS is what you want

 

[pyr02k1]

Raid 6 is Raid 5 + 1 extra parity drive. With Raid 5 you get (n-1)*GB for space, where n is the number of drives and GB is your drive size. Raid 6 is (n-2)*GB. Raid 10 is just Raid 1 (Mirror) + Raid 0 (Striping). You can Raid 10 or Raid 0+1. Some people prefer 0+1 over 10... With either case, its (n/2)*GB, as in you lose half your storage capacity. With 0+1 you mirror the striped array. With 10, you stripe the mirrored array. http://en.wikipedia.org/wiki/Nested_RAID_levels for more info on those 2.

ZFS has multiple forms of raid built in. With zfs, you can do RaidZ (Raid5), RaidZ2 (Raid6) and RaidZ3 (Raid with 3 Parity drives). Plus the typical 0, 1 and 10 I believe. With Btrfs, it's raid 0, 1 and 10 only if I'm remembering correctly. I suspect your more interested in mdadm instead of BtrFS though. I want to say mdam had raid 0, 1, 4 and 5, but I may be wrong. This is the linux implementation of software raid. With ZFS, it's built into the filesystem and all of the commands to control it come with it (when installed at least). With mdadm, it's simple enough to use, and then you can pick a file system to use on top of it. (Imagine formatting your raid 5 for ext4 or ntfs). With ZFS, your creating the array, or pool, as part of the zfs file system itself, you dont pick a separate file system.

Hopefully that helps a bit. ZFS is a good choice in most cases, and sub.mesa and a few others have tons of info floating around here on it. mdadm is good and well tested, so never a bad choice. Hardware raid has its ups and downs. The major down being the need to have a specific manufacturers set of cards if your raid card dies. With ZFS (And mdadm i believe), you can just import the data into a new system if your board dies. ZFS has a huge amount of pluses though, and you should probably go with that in my opinion. Someone else posted this yesterday and I managed to read all of it before going to bed yesterday night. http://breden.org.uk/2008/03/02/a-home-fileserver-using-zfs/ ... It isn't exactly 100% up to date, but it does have a ton of relevant information.

 

[jay2472000]

"zfs is what you want"



Sure, assuming you have the time/knowledge/flameproof suit to ask questions with it.


Theres a lot of ways of doing the exact same thing. First off, what operating systems are YOU comfortable storing and recovering data off of? Why do you think you need raid? How much data are we talking? What hardware will you use for it?