Chapter 15 DBMS-Specific Features


Index types

When you rebuild indexes, PowerDesigner determines the index type based on information contained from the table statistics. It uses the number field, which indicates the estimated number of records per table, and the percentage of distinct values to compute the number of unique values. If the user has not specified a number of rows for the table, PD assumes that the table will include at least 1 row of data.

Usually the rebuild process creates a FASTPROJECTION index for all columns. Otherwise, the following rules are applied:

Criteria Index type
If no statistics are provided and the column has an undefined data type No index is created
Low number of unique values in a column

Column used in join predicate
LOWFAST
High number of unique values in a column

No COUNT DISTINCT, SELECT DISTINCT, or GROUP BY queries required
HIGHNONGROUP
Column used in join predicate

High number of unique values in a column (more that 1000)

Anticipate COUNT DISTINCT, SELECT DISTINCT, or GROUP BY queries

Column must enforce uniqueness
HIGHGROUP
Column without numeric datatype No index is created
Column with date type DATE
Column with time type TIME
Column with datetime or smalldatetime type DTTM

For example (IQ v12.5, Table A contains 1500 rows

Column % Distinct

values
Unique

values
Rebuild indexes generates
Col_1 integer 100 1500 HG index
Col_2 integer 50 750 LF index
Col_3 integer 0 0 no index
Col_4 char (10) 100 1500 no index
Col_5 char (10) 50 750 LF index

 


Copyright (C) 2007. Sybase Inc. All rights reserved.