It would be extremely useful to see STM estimate RAID Group performance capabilities and peaks.
This would be based on the RAID group RAID type, and be calculated based on a user chosen read/write ratio and IO size, known performance profiles (templates, eg: 70% reads with medium block size, or 90% writes with 8k block size "OLTP", etc...), or (even better) dynamic values derived from observed array usage.
The value would give a "Peak" IOPS value, which can then be compared to current RAID Group performance (both as a separate graph of "estimate" peak IOPS for workload X", a % used graph of current IOPS vs peak, and overlaid as a watermark on IOPS graphs). This will give comparative meaning to RAID Group (Disk Group, Aggregate, Storage Pool, etc...) IOPS performance graphs. The estimated "Peak" could be displayed on the same graph.
"Current IOPS" will mean little to those who do not have an understanding of the estimate peak IOPS (based on workload), and this calculated peak will provide some context, even if it is a rough guide.
Additionally, this"peak" value can be used for alerts. Eg: Current IOPS on RAID Group is over 80% of calculated peak for current work-load type.
The peak and work-load type could also be graphed and reported on in itself. Eg: "Over the last 30 days we see that RAID Group XYZZY has changed from a majority of reads to a majority of writes, and the average block size has shrunk, hey, who created a SQL DB logging LUN on my media streaming disk group?!?"
The estimation feature would be for all array types and based on the known RAID type performance for the RAID group. Eg: RAID 5, RAID 10, RAID-DP for NetApp. A choice (or all types) would have to be displayed for arrays such as HP EVA, where the RAID type is based on the LUN vRAID.
Additionally, standard formula can be used, based on RAID type, disk type, disk speed, spindle count, etc... Array specific formulas can also be used, such as HP EVA SafeIOPS.