Stripe = spread the data over 2 disk without any redundancy. If 1 disk fails, all data is toast.
Mirror = write the same data on each disk. If 1 disk fails, your data can still be read from the second disk.
Raidz1 = write the data to 1 disk and a checksum block to the second disk. If 1 disk fails, the data can be recovered from the second disk. For 2 disks, this doesn't really make sense, since a mirror would be better and faster.
I would suggest you try using a mirror, but this will reduce total capacity by 50%.
