I'm also looking into this, but I *think* the way it works is along the following lines...
- 2x500GB ... straight mirror happening here. 500GB usable space
- 2x500GB + 1TB ... mirror each 500 onto the 1TB disk. 1 TB usable space. As redunancy and simplicity is the key, I dont think that it would be a case of striping the 500's and then mirroring that. Although I have been known to be wrong
- 3x500GB ... Each disk has 50% allocated to mirroring the data of one of the other disks. Yielding 3x250=750GB capacity
So in our example of...
3 500GB drives + 1 2GB drive. I am estimating that we would get 751GB of usuable space....
250+250+249+2
disk1 (500GB) | 250a | 249b |1e|
disk2 (500GB) | 250c | 249b |1d|
disk3 (500GB) | 250a | 250c |
disk4 (2GB) |1d|1e|
As I say, this is only what I *think* may be going on, and it is difficult to say with any degree of certainty, but this is the only model I can see working efficiently for the way they seem to suggest to works.
Also, they example they show in the video is misleading and does seem to suggest you can crash multiple drives, although I believe that it is just the fact that they are using syuch small data sets in the example, and they are quickly replicated across the unit.
Once again, if anyone actually
knows how this unit works, I would like to find out to be sure.
C.