From: Atri Rudra (atri@cs.washington.edu)
Date: Sun May 23 2004 - 23:03:05 PDT
An Overview of Query Optimization in Relational Systems
--------------------------------------------------------
S. Chaudhari
This survey focuses on the optimization of SQL queries in relational
databases. After a quick overview of the System R, the survey looks at
reducing the Search Space using commutativity of Operators, reducing
multi-block queries to single block and using semijoin like techniques. It
then looks at statistics and cost estimation and finally looks at
enumeration architectures in Starbust and Volcano.
(i) Are there any theoretical analysis of the consideration of interesting
orders in System R like there is for the Dynamic Programming part ?
(ii) The pulling of Join out from Join and Outer Join seems analogous to
conversion of a FO expression into a prenex normal form.
(iii) In the statistics section the survey mentions that in practice in
the absence of histograms, only the second max and second min statistics
are maintained: has there been any study about increasing the number of
statistics (say maintain the third and fourth max and min) ?
This archive was generated by hypermail 2.1.6 : Sun May 23 2004 - 23:03:06 PDT