Originally Posted by henris
One of the FlexRaid/NZFS benefits mentioned has been the ability to add drives with existing data. I'm just wondering where do these drives appear? Why didn't you add the drive to the array when you initially got it? Wouldn't the data be unprotected while you were filling the drive outside the array? Personally I don't understand this at all since the whole point of having a central storage is to have the drives there, under protection all the time. Perhaps someone can clarify the use case.
That drive will usually be a drive that used to not be protected. Many people have data on drives unprotected.
Protection comes as an after thought for many. Usually, the data is small and backup is simple and quick, which suffices.
Once the data become large and many drives are involved, that's usually when protection comes into play.
A second of benefit which has been mentioned is the ability is to take the drive and use it directly in another system. Why would you take any drive out of the protected array and risk it being damaged?
People have all kinds of reasons for wanting to take a drive out to another system such as:
- faster backup (take the drive to the backup station and back it up faster)
- for some, the data needs to be mobile
Any writes on the disk would cause a full parity re-calculation when the drive was added back to the array (or isn't this the case with FlexRaid?).
Not with RAID-F. For as long as the data is not modified, the parity remains valid.
By default, most files Windows tend to automatically edit are omitted from the RAID, which is another feature of FlexRAID (excluding certain data from the RAID).
Those files are things like thumb.db files, desktop.ini, and such.
Another cool thing with RAID-F is that when upgrading from a small drive to a larger drive, all the user has to do is copy the data from the small drive to the larger drive and not have to recompute parity (zero risk migration).
Even in the case of some system component failure, I would focus on getting the system back online rather than plugging individual drives to some other system. I at least have all my htpc's and their media server software pointing to my UnRaid array (eg. Plex Media Server and it's media libraries) and it would a total waste of time changing their configuration to match some temporary situation. In my case it's almost 100% media which is stored on the server so a bit of downtime is not critical.
I think the most pertinent argument about being able to support drive with existing data is that it is:
- and allows you leave the RAID implementation for something else very easily
With RAID-F or tRAID, users are not locked in. They can pull their drives out and that would be the end of it. No trace of RAID-F or tRAID (hum, I should rethink that - lock them in... lock them in...
Now, let's take a case where a user needs to migrate 10TB of data into a new RAID.
The user can bring his/her drives with data as-is and create the RAID under RAID-F and tRAID.
Compare that to having to have another 10TB of space somewhere where to first copy the data, format the drives, initialize the RAID, and finally copy the data into the new RAID.
Such migration would require 20TB worth of drives.
Things change. Today RAID-F is cool, tomorrow it is NZFS... and the day after that is its XYZ.
You just can't hate flexibility.