McVey explained that between 2006 and 2009, before Hadoop had officially launched, any user with their own data cluster could impersonate another by "trivially bypassing" security, adding that ...