Finally I find out 3 virtual machines which were making IOPS because of an application problem.
I earn an average of 1500 iops for one exported volume ( a csv for hyperV)
This is the evolution of iops before and after I solved the virtual machine problem
BTW, to help me finding the problem I used the local installed 3parmanagement console, which permits making reports on a specific VV, when the web-base management console doesnt offers this possiblity
result :
on exported VV side : the 3par was running with an average of 5500 iops a day , now : 3000 iops
on physical disks side : before : average of 5600 uiops and now : 5500iops
I wonder why the disk average iops doest really change
5400iops is the max theorical iops my 3par can handle (42 FC disks 1 cpg. 130iops x 42 = 5460iops )
Anyway, situation is now better regarding the 1500-2000 iops I just earned (on exported vv side). Thanks for your help !
Anyone could explain why a FC disk (max 130iops) could be displayed with 250iops in the view statpd ?