A proposal for a new way to manage SQL Statistics for data not evenly distributed. A dip in the logic of SQL Statistics!
Hi guys, I love database theory because it includes mathematics, statistics, logic, programming and data organization . I wrote this post because i read a post on linkedin about the well known " 201 bucket limitation " in the statistics used in SQL Server. Bob Ward (the Principal Architect at Microsoft, in the photo on the right while dreaming statistics and rugby...lol ) rensponded to the author of the linkedin post asking him to give a real example where this was a limitation. He also said he would verify. I then asked myself whether the logic with which statistics are managed could not be improved in any way without increasing their size. In the post on Linkedin was also referred to a post of the well-known mr. brent Ozar : The 201 Buckets Problem, Part 1: Why You Still Don’t Get Accurate Estimates Brent used the Stack Overflow 2010 database database for his examples and I will do the same in this post. If you want to follow my examples then, first of all, down...