I recently wanted to use ASM on an old single-instance server with 8 SCSI drives old-style 18-GB, all in all the server if the quarry, having 4 processors and 8 GB of RAM.
Since the server is not a front-end, and despite being equipped with a SCSI RAID controller, I also wanted to entrust the management of the ASM mirror, trusting in the efficiency of ASM to maximize performance with a multi-disc storage. Following the configuration normal redundancy, which consists of storing two copies of each extent on two different drives, re-reading the best manual I noticed a couple of nice "side effects" mirroring "intelligent" ASM.
Consider for example the case of 8 disks: If you break a record ASM automatically copy all the extents lost on other disks available, starting from the copies of these extents, resulting in an array of 7 disks ... is still normal redundancy
! And the redundancy is maintained until there is space available on the remaining disks to accommodate all the extents required to the database, you could virtually eliminate a disc at a time, waiting for the resynchronization, to only have two disks and mirroring again, or even one disc, with no redundancy, but with all the data is always available.
As corollaries we derive that, first, you can create a mirrored array simple with a number
any records (greater than 1), and second, the disc loses its meaning of "hot spare", replaced by free space remaining on each disk. The following can be seen as ASM is able to maximize performance by using more disks faster, both because of the latency throughput.
with these "toys".