How many people here are using unRAID?

[zylander]

Ive been going back and forth between different storage solutions for my home server and I am starting to really lean towards unRAID. I searched around here and unRAID is mentioned a hand full of times but no one goes into detail, its just; "I have it and I like it" or "my friend is using it and he likes it". Im just looking for some more detailed reviews from people using unRAID, things like;

-How many drives are in your array?
-What other storage solutions (if any) have you used? (WHS, FreeNAS, RAID, software RAID) And how does unRAID compare to those other solutions?
-What are your hardware specs?
-What version are you using? Basic, Plus or Pro?

I really just want to hear some reviews form people who have been using it regularly.

 

[taltamir]

unRAID is basically multiple drives in a complex and intelligent system that removes a lot of the faults of RAID...
Freenas and the other similar ones are a basic linux install with a GUI geared specifically for controlling a NAS.
I use ZFS, which removes a ton of the limitations of RAID, and has really awesome capabilities and top notch data safety, and it is free and open source.

 

[Emulex]

jbod is alot more reliable when using consumer class drives.

i spose you could just use rsync to do a similar task but whs is easy to acquire.

the commercial product is microsoft storage server with single instance deduplication/dpm (compared to WHS)

 

[RebateMonger]

Originally posted by: Emulex
jbod is alot more reliable when using consumer class drives.

Just don't use the (dynamic disk) Windows Software version of JBOD and expect it to be reliable with more than a couple of drives. If ANY of the spanned disks fails, you lose ALL the files on ALL the disks. I'm not saying the files on the "good disks" couldn't be recovered, but you can't just replace a single bad drive and "rebuild" the JBOD. You have to start fresh, build a new emtpy spanned array, and restore all your files from backups.

Of course, as long as you have backups, it's fixable, but a Windows software JBOD array has got to be the least reliable version of all the Windows software arrays. A RAID 0 array will typically only have two disks to fail, but a JBOD array may have several, so the odds of disk failure are much higher.

 

[taltamir]

JBOD is crap, it is the most unsafe storage form around, you are much better off just having the drives as individual disks.
ZFS > unRAID > RAID 10 > RAID1 > RAID6 > RAID5 > individual disks > JBOD

 

[Blain]

Originally posted by: taltamir
ZFS > unRAID > RAID 10 > RAID1 > RAID6 > RAID5 > individual disks > JBOD

RAID 6 :thumbsup::laugh:

 

[Emulex]

no i'm saying consumer (non-tler) large drives are better off with a SIS/rsync jbod than with any raid.

the raid controllers freak the heck out when you hit a 30 second remap.

TLER sata and enterprise SAS is another story; most of us do not use that setup.

openfiler self destructed with a pair of brand-new-refurb 750gb seagate's when they spent their pretty time trying to fix a soft-defect.

raid-6 has insanely poor disk write speed in soft-raid.

 

[zylander]

What really drew me to unRAID was the option to have drives not in use spin down and the ability to combine drives into what they call user shares so that to client computers it gives the illusion of one disk like RAID.

I have not heard of ZFS but Im reading up on it. Does it have the option to combine multiple directories into one like unRAID?

 

[taltamir]

Originally posted by: Blain

Originally posted by: taltamir
ZFS > unRAID > RAID 10 > RAID1 > RAID6 > RAID5 > individual disks > JBOD

RAID 6 :thumbsup::laugh:

yea, it rocks... ZFS can use all those raid forms... I use ZFS RAID6. (raidz2)

Originally posted by: Emulex
no i'm saying consumer (non-tler) large drives are better off with a SIS/rsync jbod than with any raid.

the raid controllers freak the heck out when you hit a 30 second remap.

TLER sata and enterprise SAS is another story; most of us do not use that setup.

openfiler self destructed with a pair of brand-new-refurb 750gb seagate's when they spent their pretty time trying to fix a soft-defect.

raid-6 has insanely poor disk write speed in soft-raid.

Early versions of solaris got me 30MB/s... v117 gives me up to 80 MB/s write. it is not as fast as my intel SSD. but it is much faster than any single spindle drive I ever owned, and reads are excellent.
I am running 4GB of DDR2, and an amd low power 2.4ghz Athlon64 X2. and 5 x 750GB WD green power drives @ 5400 RPM (despite their poor writes, they are cheap).

the advantage is data integrity and cost. If i needed more speed on my fileserver I'd do RAID10 in ZFS.

 

[zylander]

So does this ZFS installed as its own OS, or can it be run along side another OS like Win Server or linux?

 

[taltamir]

ZFS is a filesystem developed by sun, originally for their own operating system, solaris. Which they open sourced...
Apple has ported it to mac OSX server (I do not see why not give it to regular customers since it is free and open source!), and it was also ported to BSD. However linux license forbids itself from being linked with certain other licenses, including the one ZFS is under.
As a result linux requires some special acrobatics to do, either port it under fuse, rewrite the code to match the standard, etc... so linux porting is not going well...

That being said, linux is actually only a KERNEL without a full operating system. They run GNU/Linux... well, people said... why not run GNU/Solaris? So you have several projects doing just that... including the major official one called opensolaris.

Go to genunix.org for a good source to download solaris kernel derivatives, including the main opensolaris

Anyways, the most up to date and most well developed version is the solaris one... so that is what I use for my fileserver. You can easily transfer though... A quick reformat of my solaris system with BSD, and I just type # zpool import -f tank
And it imports the array with all the data... same with other OS versions. I actually tried it before putting any data on...