I have the ff setup:

Solaris 8
DiskSuite 4.2.1

1) Active database on a Sun Enterprise 3500 with dual A1000 disk arrays.
Disks inside the array are stripped using hardware RAID 0 in the A1000.
Then one A1000 is a mirror of the other, for RAID 1 mirroring using Solstice DiskSuite.
Each array connected to separate SCSI controllers ( QLogic PCI-SCSI cards ).
This setup has saved us a couple of times, when disks in the array would be reported as "Needs Maintenance" by DiskSuite,
because the other half of the mirror in the other array is still working.

Questions:
How does Solstice DiskSuite, and Solaris itself, behave in the event of :
1a) a failure on the SCSI controller ( the QLogic PCI-SCSI card )
1b) a failure on the A1000 array itself, e.g. backplane, etc.

Will DiskSuite still keep working in that, it will just ignore reading and writing to the failed array ?
Or will it timeout, crash, panic ?



2) Standby database on a 280R with dual 3310 disk arrays.
Now currently debating with the disk arrays in the 3310 should be configured similar to the A100s .. OR
go RAID 5 on each array, and leave one disk on each array as a global hot spare.
This also allows us with a lot of free disk space.

I guess the deciding factor whether I should go RAID 5 vs. RAID 0 + 1 will rely on the answer
to my DiskSuite questions [1a] and [1b].

If DiskSuite crashes, timeouts, panic, etc ... then there is really no point with the mirroring between
two arrays with DiskSuite anyway.



Any thoughts / suggestions ?

Does RAID 5 hot spares replacing failed drivers work well with raw partitions ?

Thanks