I have a server that I built about 2.5 years ago:

 

Norco 4224

Supermicro board + Xeon processor

16gig ECC ram

Areca 188X (can't recall) + 2gig ram

1x10 drive raid 6 array and 1x8 drive raid 6 array

WHS

 

I'm helping a friend with an unRAID build right now mainly because it is cheaper than my route with Areca card and should require less frequent hands on maintenance once up and running.

 

As I'm doing this I'm trying to determine if there is any benefit to me to ditch WHS and move to unRAID or, given that I dumped a grand on a raid card, not really.  Could you guys provide some feedback on the benefits of unRAID to WHS and what the implications to me from a RAID setup / RAID card perspective might be?

 

Server runs:

SABNZBD

Couch Potato

Sonarr

MySQL

Plex

 

Thanks.

Given that you've already got a very nice server with RAID-6 arrays (thus dual fault tolerance), it's difficult to claim that UnRAID would be an improvement.

 

There are a couple of "benefits" ... but they come at a cost in overall reliability.

 

(a)  You'd have more storage in an UnRAID array with the same set of disks ... since instead of having 4 disks dedicated to providing fault tolerance (2 for each of your arrays), you'd only have one => so the other 3 would provide additional data storage space.    The "cost" of this is, of course, far less fault tolerance.

 

(b)  In the event of a catastrophic failure (where you lose more drives than you can tolerate for fault tolerance), UnRAID will result in less data loss, since only the data on the failed drives is at risk.  Of course with RAID-6 arrays, this would requires THREE simultaneous failures, whereas UnRAID only requires TWO.

 

©  In general, your drives won't be spun up as much, since UnRAID only needs to spin up the drives actually being read (since the data's not striped).    This also, of course, means the performance is lower on UnRAID.

 

 

Other than what gary has said above, I migrated from Unraid to WHS 2011 then back again.

 

Granted I haven't got a RAID card like yours and that bit you need to work out for yourself, only you know what the case scenarios above would mean to you.

 

On the subject of the apps you've posted all I can say is it is far easier to get them all setup in a docker environment in Unraid than it ever was in WHS.

 

Would I move back to WHS 2011 now I've come back to Unraid - A resounding no, I can accomplish things I wish my server to do far easier, quicker and with less maintenance and also achieve some things that were never really satisfactory with WHS

 

For example I had a cludge of a fix with WHS in getting my  kodi library centralised and auto-updating.

 

Mysql installed fine, but I had to alter the registry to get WHS to auto login at boot and run kodi, leaving administrator logged in permanently.  This took an awful lot of work and head scratching,  in Unraid - there's a docker for that and job done in about five minutes.

Agree the Dockers are a significant advantage of UnRAID over WHS I should have mentioned.  I didn't because the OP already has the applications he wants setup and configured in his server.

 

I answered primarily from a NAS perspective => UnRAID is an excellent NAS, but it's certainly not better than a pair of RAID-6 arrays.  The RAID-6 arrays will be appreciably faster and are fault tolerant for two failures.    The only real advantage UnRAID has relative to those is the ability to mix and match drive sizes ... which is unlikely to be a factor in this case.

 

Don't get me wrong ... as I'm sure everyone knows, I'm a BIG fan of UnRAID.    But I've also set up a few RAID-6 cards for folks ... when they want dual fault tolerance that's a far better choice => and I DO like the dramatic speed increase they get relative to my UnRAID arrays.

 

(b)  In the event of a catastrophic failure (where you lose more drives than you can tolerate for fault tolerance), UnRAID will result in less data loss, since only the data on the failed drives is at risk.  Of course with RAID-6 arrays, this would requires THREE simultaneous failures, whereas UnRAID only requires TWO.

One other minor point that goes along with this advantage is that each disk has it's own file system which (b) implies but doesn't specifically state.  That is what I consider the biggest advantage and why I will not consider a fault tolerant option that doesn't include it.  For me it doesn't matter how many drives I have for fault tolerance as long as the data drives are not stripped it is good enough for me.  Some how I would find a way to have at least 1 more drive then I have a fault tolerance for fail at the same time - just they way my luck runs.  But with unRAID even with those failed drives I can usually recover 95-100% of the files that were on it - can't do that with a stripped array.

One other minor point that goes along with this advantage is that each disk has it's own file system which (b) implies but doesn't specifically state.  That is what I consider the biggest advantage and why I will not consider a fault tolerant option that doesn't include it.  For me it doesn't matter how many drives I have for fault tolerance as long as the data drives are not stripped it is good enough for me.  Some how I would find a way to have at least 1 more drive then I have a fault tolerance for fail at the same time - just they way my luck runs.  But with unRAID even with those failed drives I can usually recover 95-100% of the files that were on it - can't do that with a stripped array.

yes, the usable filesystem per disk is a huge benefit of unRAID and should be called out. My current storage has this as well, but very differently. 

 

In my old job I routinely (way too often) did recovery on multiple disk failure on arrays with striped data. And like you, could recover all the data, or identify the few unrecoverable blocks on specific files. The trick is getting/keeping the array online during the filesystem work. Most of the time I worked on software raid arrays with dual parity which makes keeping the array online much easier since it is just minor changes to the error handling code. Unfortunately due to the size of the arrays, the recovery process could run for months.

 

The problem I have with home use of hardware raid, what happens when the controller dies? For me that was often well after they vanished from store inventory, replaced by newer better (incompatible) models. Thus I only recommend software raid for home use. I am sitting on a pile of aging 3ware and acera controllers "just in case" they are called for duty one last time.

An advantage of unRaid is the ability to incrementally grow the array with drives of different sizes. This is a big one for me. I have no idea how you'd grow a Raid array without enough disk space to unload the entire array and completely reload it with data. With unRaid I can replace a disk with a larger one or add a new disk with minimal disruption.

 

Raid puts a higher value on data integrity. For example, when unRaid red balls a disk as a file is being copied, that file will get corrupted on the simulated disk. With Raid no corruption would occur. In use cases where data corruption could or would occur, the preference of Raid arrays is to fail the entire array, and the preference of unRaid is to minimize data loss.

 

Raid arrays will have higher performance in general.

An advantage of unRaid is the ability to incrementally grow the array with drives of different sizes. This is a big one for me. I have no idea how you'd grow a Raid array without enough disk space to unload the entire array and completely reload it with data. With unRaid I can replace a disk with a larger one or add a new disk with minimal disruption.

Arrays grow by adding disk as well, but typically same size drives (or the excess is lost). This is true of hardware and software RAID arrays. They also allow for change of RAID level. Most filesystems also support this, XFS even requiring to be online

... An advantage of unRaid is the ability to incrementally grow the array with drives of different sizes.

 

The only real difference between UnRAID and a typical RAID-5 or RAID-6 is the "... of different sizes" part.  And while that was a neat feature 9 years or so ago when UnRAID was new, drive prices are so much lower these days that it's not really much of a burden to require all drives to be the same size.    If/when you want to rebuild with larger drives, you can simply repopulate the new array from your backups.

 

 

... Raid puts a higher value on data integrity.

 

I wouldn't really say there's any difference in the focus on integrity.  A file being copied to a RAID-5 array that fails mid-copy can also be corrupted.  But except for failure-during-a-write, both are very good at preserving data integrity.

 

 

... Raid arrays will have higher performance in general.

 

No "in general" about it => a RAID-5 or RAID-6 array with a significant number of disks will blow-away the performance of an UnRAID setup.  By an order of magnitude or even better !!  I've seen an Areca RAID-6 setup with a dozen disks that had write speeds over 450MB/s and reads in the 800MB/s range.    And this is with dual failure fault tolerance !!

 

... An advantage of unRaid is the ability to incrementally grow the array with drives of different sizes.

 

The only real difference between UnRAID and a typical RAID-5 or RAID-6 is the "... of different sizes" part.  And while that was a neat feature 9 years or so ago when UnRAID was new, drive prices are so much lower these days that it's not really much of a burden to require all drives to be the same size.    If/when you want to rebuild with larger drives, you can simply repopulate the new array from your backups.

 

Depends on your array size. If a user has, say, 50T of data, and wants to move from 19 3T drives to 10 6T drives, that would cost a person $2500 in new drives (assuming $250/6T drive). I don't agree that is "really not much of a burden".

 

... Raid puts a higher value on data integrity.

 

I wouldn't really say there's any difference in the focus on integrity.  A file being copied to a RAID-5 array that fails mid-copy can also be corrupted.  But except for failure-during-a-write, both are very good at preserving data integrity.

 

Really? I think RAID5 or RAID6 would fail-over without data corruption. If not, corporate IT would have a cow!

 

... Raid arrays will have higher performance in general.

 

No "in general" about it => a RAID-5 or RAID-6 array with a significant number of disks will blow-away the performance of an UnRAID setup.  By an order of magnitude or even better !!  I've seen an Areca RAID-6 setup with a dozen disks that had write speeds over 450MB/s and reads in the 800MB/s range.    And this is with dual failure fault tolerance !!

 

You can always come up with a weird use case, but agree. RAID is the way to go for high performance.

... An advantage of unRaid is the ability to incrementally grow the array with drives of different sizes.

 

The only real difference between UnRAID and a typical RAID-5 or RAID-6 is the "... of different sizes" part.  And while that was a neat feature 9 years or so ago when UnRAID was new, drive prices are so much lower these days that it's not really much of a burden to require all drives to be the same size.    If/when you want to rebuild with larger drives, you can simply repopulate the new array from your backups.

In my view this is a HUGE plus point as a typical home user is likely to have accumulated their drives over a period of time and they are thus very likely to be of mixed size.  At this point this is a sunk cost so being forced to replace ALL drives when you want to start using a larger drive size is a large barrier to moving to larger size disks.

 

 

... Raid puts a higher value on data integrity.

 

I wouldn't really say there's any difference in the focus on integrity.  A file being copied to a RAID-5 array that fails mid-copy can also be corrupted.  But except for failure-during-a-write, both are very good at preserving data integrity.

This is true at the file level as both do not provide protection against a sector going bad silently corrupting data, but not true at the total system level.  Typical RAID systems do not have each drive as a discrete file system and thus usable outside the array.  I have been bitten by this in the past where I have had a failure on an RAID-5 system that resulted in total loss of data, whereas on unRAID in a similar scenario I was able to do recovery at the individual disk level leading to almost total recovery of data. 

 

... An advantage of unRaid is the ability to incrementally grow the array with drives of different sizes.

 

The only real difference between UnRAID and a typical RAID-5 or RAID-6 is the "... of different sizes" part.  And while that was a neat feature 9 years or so ago when UnRAID was new, drive prices are so much lower these days that it's not really much of a burden to require all drives to be the same size.    If/when you want to rebuild with larger drives, you can simply repopulate the new array from your backups.

 

In my view this is a HUGE plus point as a typical home user is likely to have accumulated their drives over a period of time and they are thus very likely to be of mixed size.  At this point this is a sunk cost so being forced to replace ALL drives when you want to start using a larger drive size is a large barrier to moving to larger size disks.

 

I suppose that's true ... although virtually any modern controller can support multiple arrays, so it's simple to just start a 2nd array with larger drive sizes => and as you buy more drives just increase the size of the newer array and decrease the drive count of the older one.    But I'll grant that this is a nice feature relative to RAID-5 or RAID-6 arrays.

 

 

... Raid puts a higher value on data integrity.

 

I wouldn't really say there's any difference in the focus on integrity.  A file being copied to a RAID-5 array that fails mid-copy can also be corrupted.  But except for failure-during-a-write, both are very good at preserving data integrity.

 

This is true at the file level as both do not provide protection against a sector going bad silently corrupting data, but not true at the total system level.  Typical RAID systems do not have each drive as a discrete file system and thus usable outside the array.  I have been bitten by this in the past where I have had a failure on an RAID-5 system that resulted in total loss of data, whereas on unRAID in a similar scenario I was able to do recovery at the individual disk level leading to almost total recovery of data.

 

Agree that the total data loss in a multiple failure case with RAID-5 is a major disadvantage vis-à-vis UnRAID.    But I'd NEVER build a RAID-5 array these days ... there's no reason not to use RAID-6 => and that problem is FAR less likely with dual fault tolerance.

 

In any event, regardless of whether you're using UnRAID, RAID-5, RAID-6, or any other RAID configuration, you should always have current backups, so just how much data you'd lose in some catastrophic failure situation should be the same:  NONE