Database Engineering and Applications Symposium, International (2000)
Sept. 18, 2000 to Sept. 20, 2000
B.J. Oommen , Sch. of Comput. Sci., Carleton Univ., Ottawa, Ont., Canada
M. Thiyagarajah , Sch. of Comput. Sci., Carleton Univ., Ottawa, Ont., Canada
Histogram techniques are used to efficiently estimate query result sizes in most of the modern-day database systems. In a recent work (Oommen and Thiyagarajah, 1999), we introduced a new histogram-like approximation strategy, called the Rectangular Attribute Cardinality Map (R-ACM), which approximates the density function within a given sector by a rectangular cell. In this paper, we introduce another histogram-like approximation strategy, called the Trapezoidal Attribute Cardinality Map (T-ACM) that approximates the density function within a given sector by a trapezoidal cell, where the slope of the trapezoid is obtained so as to fix the actual probability mass within the cell. We present numerous analytic and experimental results concerning the T-ACM demonstrating its superiority over the traditional equi-width and equi-depth histograms for query result size estimation. We hope that with the R-ACM introduced in (Oommen and Thiyagarajah, 1999), the T-ACM could become an invaluable tool for query optimization in the future database systems.
database theory; query result size estimation; Trapezoidal Attribute Cardinality Map; query optimization; histogram-like approximation strategy; Rectangular Attribute Cardinality Map; density function approximation; probability; experimental results; T-ACM; equi-width histograms; equi-depth histograms; R-ACM
B.J. Oommen, M. Thiyagarajah, "Query Result Size Estimation Using the Trapezoidal Attribute Cardinality Map", Database Engineering and Applications Symposium, International, vol. 00, no. , pp. 236, 2000, doi:10.1109/IDEAS.2000.880582