alf » from archive
RT @brendangregg: casual benchmarking: you benchmark A, but actually measure B, and conclude you've measured C.