Meaning of data from "onstat -g iof" under 7.13

I need to optimise the placement of my tables/indices within my dbspaces and the placement of my dbspaces on my physical disks. The setup is 32 IBM SSA disks, each pair mirrored by the host, an RS6000 running AIX 3.2.5 with online 7.13.UC3. To do this I have tested several sources of data: 1. The contents is the SMI table sysptprof 2. The output of the "onstat -D" command 3. The output of the "onstat -g iof" Unfortunately, while "onstat -D" and the SMI tables agree (most of the time), the "onstat -g iof" command reports conflicting information. In fact the "onstat -D" output for my most busy dbspace (according to "onstat -g iof") reported 55,000 reads last night, 48,000 reads this morning and 300,000 reads this lunchtime, i.e. the number of reads actually when DOWN overnight. "onstat -" reports that the server has been up for 54 days - could it be that "onstat -D" figures are only the least significant bytes of the real value? I am certain there have been more than 48,000 reads on my hottest table during the last 54 days! My gut feel tells me that "onstat -g iof" is correct. However, the manuals don't contain any information on exactly what the values it produces mean. In particular does anybody know what a sensible value for IO/s might be on 80Mb SSA or 40Mb Ultra SCSI disk? Is this an average or instantaneous value? Would it be artificially low if the server was inactive for 4 hours over night? Any assistance or advice would be greatly appreciated. Graham Single grahams@itl.net