If you really want to baseline things and isolate the changes made, ideally, you might do something like record a few segments (hours worth, including all aspects - input, output, change, daily activities, weekly activities, maintenance, etc.) of your normal workload, and then replay it on a freshly rebooted, completely isolated physical instance that was (pre-reboot) restored to a point in time just before the segment was recorded.

Record the various counters (sys.dm_os_performance_counters, sys.dm_io_virtual_file_stats, perhaps even sys.dm_db_index_usage_stats or a Perfmon trace to get Disk Avg Sec/Read and Disk Avg Sec/write) over a few restore, reboot, replay cycles. Make your changes, then do another few cycles, and look at the overall changes.