N00b Build & questions

[Wozzzaaa]

Hi All
This is my first Freenas build & i ve got some questions. I hope some kind soul can help me. Ive got a lot of Windows experience but fairly average Linux skills. Ive never really played around with NAS & RAID before. This is what ive planned from research.

Ive got a Fractal R3 Case with 8 x HDD Bays. What I want is the following

1) Critical Data

Storage for critical data family photos / home movies. Must be able to survive total failure of 1 drive. I was thinking 2 x 1tb drives in a 1 mirror.

2) Non critical Data
Stuff that if lost, I could live without, however it would be an inconvenience to replace.

Ive got 6 bays free, and have the following Drives

5 x 2tb
1 x 1tb

What would be the best way to pool these drives? I would like to have storage space over redundancy. If I use a RAIDZ, the storage will be lowered due to the 1tb drive?

Thanks

 

[DoS]

well if you want performance & redundancy, then that calls for raid0 right there.

just for referance:

raid0: most performance, no capacity loss, no redundancy.
raid1: improved performance, 50% storage loss, most redundancy.
raid0+1: previous two combined. 4 drives needed w/ 50% capacity loss.
raid5: ok performance, n*x-n capacity loss, single drive failure
raid6: ok performance, n*x-2n capacity loss, two drive failure.
raidz: a software implementation of above. benefits over hw raid are theoretical.
(n=capacity, x=#drives)

having the single 1tb drive does lead to having an unbalanced raid. seems like you would be well off having the 1tb drive on its own, the the 5 2tb in a raid5/ raidz config. this would give some redundancy in the instance one drive in the array fails. you could create two pools and lump them together as well. so p1 = 1tb, p2 = 2tb*5 (depending on raid), then add p1 & p2 to make one large volume. but i believe if that single 1tb drive fails, you would wind up losing everything this way. the p1 & p2 arrays are essentially striped/extended together.

storage rule of thumb: there are only drives that have failed, and drives that are going to fail...

 

[SoftDux-Rudi]

Rather use ZFS and setup a stripped pool for the critical data, and raidz1 for the non-critical data. ZFS offers better redundancy than RAI0 & RAID5

 

[matthewowen01]

you could create two pools and lump them together as well. so p1 = 1tb, p2 = 2tb*5 (depending on raid), then add p1 & p2 to make one large volume. but i believe if that single 1tb drive fails, you would wind up losing everything this way. the p1 & p2 arrays are essentially striped/extended together.

this is not a good idea as you don't get to choose which vdev the data is written too, ZFS will distribute the reads and the writes according to load and will be distribute the data across the 2 vdevs. then if you loose the 1 TB drive, all is lost.

Rather use ZFS and setup a stripped pool for the critical data, and raidz1 for the non-critical data. ZFS offers better redundancy than RAI0 & RAID5

you are mistaken on the terminology. stripping has no redundancy because the data is stripped across all the drives. Raid0 is not a true raid in the sense of redundancy data is evenly distributed across all drives without any redundancy. raid1 is mirroring, which is what i think you are thinking of.


the best bet is to create a raidz2 of the 5 2TB drives (or possibly buy a 6th one) and ignore the 1 TB drive. you'll have 6 TB (or 8) of usable space with 2 disks of redundancy. the 1 TB drive is just a waste of a SATA port.

 

[SoftDux-Rudi]

you are mistaken on the terminology. stripping has no redundancy because the data is stripped across all the drives. Raid0 is not a true raid in the sense of redundancy data is evenly distributed across all drives without any redundancy. raid1 is mirroring, which is what i think you are thinking of.


the best bet is to create a raidz2 of the 5 2TB drives (or possibly buy a 6th one) and ignore the 1 TB drive. you'll have 6 TB (or 8) of usable space with 2 disks of redundancy. the 1 TB drive is just a waste of a SATA port.

Sorry, that's what I meant:


Rather use ZFS and setup a mirrored pool for the critical data.