Week 10 Lecture (9)

9

Exercise 1: Simple Hash Join Cost

Consider executing Join[i=j](R,S) with the following parameters:

r_R = 1000, b_R = 50, r_S = 3000, b_S = 150, c_Res = 30
R.i is primary key, each R tuple joins with 2 S tuples
DBMS has N = 43 buffers available for the join
data + hash have reasonably uniform distribution

Calculate the cost for evaluating the above join

using simple hash join
compute #pages read/written
compute #join-condition checks performed
assume that hash table has L=0.75 for each partition