I am not sure if I can ask about SPM here. I ran into a situation that if FlashCache is specified then the throughput prediction goes down.
FAS2650 comes with builtin 1TB FlashCache. My config is one FAS2650 with 2 DS224C fully populated with 24 960SSD's, and each FAS2650 controlloer owns one shelf, thus one aggregate, respectively. Then I generate 2 workload profiles of 5TB capacity, 90% random read, 10% random write. The 2 workloads are assigned to 2 aggregates of 2 controllers respectively.
If I chose no FlashCache, the overall throughput is 120K IOPS, and if I chose 512GB FlashCache per controller, it turned out 113K IOPS. Is this reasonable?