The above logic says that query scheduler thread executes start at very second and checks to see which queries are eligible to run that time based upon the splayed_interval( if (query.splayed_interval > 0 && i % query.splayed_interval == 0) ). Now if we dont apply schedule_splay_percent then scheduler will try to execute every query with same interval at the same time. By applying schedule_splay_percent we are giving scheduler the freedom to execute the queries with same interval in a time range.