What a terrible blog post. It basically ends with “well, this test is meaningless under real world conditions.”

How about some more robust testing that mimics real world scenarios?