Just confirming that on a non-cluster node w/proper AUTOGEN tuning that I don't need to worry about large lock rates that appear in concert with large direct I/O rates.