IBM 000-8697 User Manual

2-138

IBM Informix OnLine Database Server Administrator’s Guide

Structure of an Index Page

Figure 2-34

illustrates the root node page and the two leaf node pages that

result from a split after the root node fills.

Index Key Entries

Figure 2-34

includes index key entries on the root node index page that take

the following form:

■

A byte value followed by one address of a branch or leaf node page

■

Only a node address (the infinity slot)

In addition, a third form is possible:

■

A byte value followed by a rowid, followed by two node addresses
(indicating a range of pages)

These three types of index key entries are described in the paragraphs that
follow. The entry types are illustrated in

Figure 2-35 on page 2-139

When the byte value is followed by a single branch or leaf node address, the
index key entry indicates that only one rowid exists for this byte value. The
byte-address pair entry points to the first data slot on the node page specified
by the address. The node page can be either a leaf node page or a branch node
page.

Figure 2-34

After the root node

page fills, it splits

into two leaf nodes.

The infinity slot

points to the node

that contains all

byte values greater

than the last value

actually stored at

this level. In this

example, the

infinity slot points

to all values greater

than O’Brien.

Key values from Albertson to Miller,
arranged in byte:rowid entries

Key values from O’Brien to Watson,
arranged in byte:rowid entries

O’Brien: leaf node address

Page header

Infinity slot

Albertson: leaf node address

Header

Root node

Leaf nodes

Horizontal

Link