The Case for Learned Index Structures

   Abstract