I have done the second item on your list and the disc queue is always full during use, and processor is at about 25%. From this I guess my bottleneck is in disc bandwidth. Is this correct, or should I check other things too?
I have no clue about the first item, checking wait events. Is this information got from v$sysstats?
1. Looks like you are right, but also check how load is distributed
among your disks
2. v$system_event, v$waitstat, v$sysstat, v$sesstat but better use such
tools as oracle performance diagnostic tool - STATSPACK or simple
bstat/estat. You can also use other performance diagnostic tools.
I guess this is the problem. It is recommended to avoid RAID 5 on systems with high volume of inserts/updates. I think you should consider
the possibility to reorganize your disks in RAID 10. It may greatly alleviate the problem. And if you do this do not forget to choose the correct stripe-width value.
RAID 10 is supposed to provide the better performance for I/O intensive applications (especially for inserts/updates) than RAID 5. In your case it is likely that you will get a sufficient performance gain.
As to the external raid - consider SPARCstorage Arrays, it may be integrated with different platforms and OS.