NFS has several operation types that are not purely Reads or Writes. Checks of last access time, or checks of attributes are recorded as an NFS "Other Op". Some NFS workloads are very heavy in other type ops.
These ops are usually fairly light and have little impact on the overall system load. For this reason its useful to separate them out from the read and write traffic. In one situation at my customer site we had a FAS960 doing 25000-30000 NFS ops a second, which is an implausible number for that model of FAS appliance. When looking more closely at the traffic, however, the vast majority of the operations were "other" ops. The workload had a very high file count and the application was designed in such a way that many getaddr type operations were required.
You won't find "system" ops recorded under any of the protocol types. If you are looking for operations that OnTap is using for its "under the covers" work like dedupe or block reclamation its likely you will need to run perfstat.
You can get a sense of whether the filer is running a lot of "under the covers" by looking at CPU load vis-a-vis the protocol workloads. If there are very few CIFS/NFS/FCP/ISCSI operations a second, but the CPUs are very high, then perhaps you have a high load of backend work on the appliances. Drawing that conclusion without running a perfstat would be conjecture though.
That means, Does the "Other_ops" include only client I/Os ? My customer want to know below, Does "Total Ops/Sec" in Operations Mangers(not Performance Advisor!) about volumes shows IOPS form clients ?
I think that above "Total Ops/Sec" is same as "total_ops" of "Volume IOPs View" in Performance Advisor. And I saw in my test environment, SnapMirror had started and "Other_ops" in Performance Advisor was grown. So, I think that the "Other_ops" includes except for clients access too.
Therefore, the "Total Ops/Sec" in Operations Manager(and "total_ops" in Performance Advisor) shows not only from clients I/O but also all I/O for the volume.