DSM-G600, DNS-3xx and NSA-220 Hack Forum

It is not really a matter of opinion, just fact.

It is important that you understand RAID 1 and what it is used for.  RAID 1 makes a real-time mirror of your disk.  For example, with 2 1TB drives, you will only have 1TB of usable space.  The purpose of RAID 1 is to keep the system up and running after a single physical hard drive failure.  This should mean access to the unit remains the same as before the failure.  This should be invisable to the end user.  The theory is that the drive then gets replaced, and then after the new drive fully syncs, everything is back to normal running.  As I said, this is completely invisable to the end user.

The disadvantage to this is that you still require backups of the system.  If a user erases something, or a file is corrupted, or a virus strikes, it does so to both disks.  There is no backup of the system.

By copying 1 drive to the other,  you now have incremental backup of the system.  Anything changed on the main drive is not changed on the second drive until the copy is made.  In this instance, if a user or virus or whatever erases or destroys data, it is only destroyed on the primary drive, not the second drive, unless you discover it after the copy was made. 

The disadvantage to this is that if the primary drive fails, the system will be down until you either rebuild it, or switch your backup over to be the main drive.  Also, any work that was done after the last copy, but before the failure will be lost.

It is really up to you to determine the value of your data and your time.  You need to ask yourself if your data needs to be accessable all the time, even in the event of a drive failure.  Either way you go, it is ALWAYS prudent to have a backup of your data, especially of your non-replacable data.

Now for my opinion:

There have been some issues with DLink's implementation of RAID 1 on the DNS.  It is manily bugs in their firmware which has resulted in some incorrect formats (wrong drive) and hang-ups during configuration.  This is not an issue with the software RAID, only with DLink's interface to it.  The RAID setup works very well an is highly reliable if you choose to learn how to use mdadm and configure things manually.  It does take some Linux knowledge and there are a few files that need to be changed and updated while doing it, but it works very well.  I have done a lot of testing with this. 

I personally use individual drives and back them up via the USB port to external drives.  I do not have a need to have the system up all the time, and I also have scripts made so I can swap out a bad drive to a backup very quickly.  I also do not store a lot of live data on either of my DNSs.  I use them mainly for serving movies to my 3 Popcorn Hour NMTs.

Once again, the question about if Raid 1 is what you need or not is a matter of if you need the data available 24/7 even in the event of a drive failure, or would you rather have the storage space?  The idea of inserting a new disk and everything is fine is great in theory, but not so great in practice.  There are a few things that you would have to do.  Now, the next question I will ask is if you need the DNS' web admin tool to be happy?  This adds some complexity.  If you have a basic understanding of Linux, putting in a new disk, partitioning it, formatting it, and syncing it to the array are pretty easy to do.  The unit will work by doing only these things, but there are config files that the web admin tool looks at each time you log in that would not be right.  Due to this, it would ask you to format each time you log in (of course hitting cancel will take you to the rest of the web admin tool and it will work no problem).  The config files reside in 4 location: 1 on each disk and 2 in flash.  The files contain the information as to the raid (or not) configuration, as well as the physical serial numbers of the hard drives.  All these files need to be updated each time a new drive is put in.  The only advantage in this respect of the individual drives is that not all of the config files are located on each of the drives in individual mode, saving a small amount of work.

The unit has also been know to format the wrong drive even in individual mode, so I always do everything manually.

Lastly, my backup drives are the same make and model as the internals.  I also set up the partitions the same.  This way, if I get a failure, I can drop the backup drive right in the unit, copy a couple of files to flash, and I'm back up and running in 5 minutes.  I then just start making another backup. 

You really need to make the decision yourself between the cost of having the redundancy (disk price) and the cost of time to restore a backup.  For me is is more cost effective to have the larger amount of storage and restore as I stated above.