---

Larger example of join estimation

---

• Example with relations with more attributes
  • Statistics on the input relations:

    R(a,b,c) S(b,c,d) U(b,e) ======================================================= T(R) = 1000 T(S) = 2000 T(U) = 5000 a V(R,a) = 100 b V(R,b) = 20 V(S,b) = 50 V(U,b) = 200 c V(R,c) = 200 V(S,c) = 100 d V(S,d) = 400 e V(U,e) = 500

    Problem:

    Estimate the size of R ⋈ S ⋈ U

    ---

    ---

  • Solution:

    T(R) × T(S) T( R(a,b,c) ⋈ S(b,c,d) ) ~= -------------------------------------------- max( V(R,b),V(S,b) ) × max( V(R,c),V(S,c) ) 1000 × 2000 ~= ---------------------------- max(20,50) × max(200,100) 1000 × 2000 ~= ------------- 50 × 200 ~= 200 T( R⋈S ) × T(U) T( (R⋈S) ⋈ U ) ~= ------------------------ // R⋈S and U has attr b in common max( V(R⋈S,b) × V(U,b) ) 200 × 5000 ~= -------------- // V(R⋈S,b) = V(S,b) = 50 max(50, 200) ~= 5000

    You can try different ordering and see if you get the same answer....