Enscribe Programmer's Guide

Typically, in a changing database most blocks will be approximately two-thirds full at any given

time. When using the FUP LOAD command to load data into a key-sequenced file, you can specify

how much empty space to provide for future growth.

Figure 14 Key-Sequenced B-Tree Structure

AMY

CAROL

LOUISE

First (Highest)-Level Index Block

Second-Level Index Block

Data Blocks

Second-Level Index Block

0 MOLLY VINCE

0 CAROL LOUISE

ytpmEYTTEB

ytpmESIRHC

ytpmEYLLIM

MOLLY OLGA STEVE

ytpmE ETANYLLOM

ytpmEYTSURAGLO

ytpmEANELUEVETS

Data Record

Primary Key

Variable-Length

Physical Records

Saved

for Future

Insertions

NATE is alphabetically greater than MOLLY but less than VINCE.

Go to the second-level index block that begins with MOLLY.

NATE is alphabetically greater than MOLLY but less than OLGA.

Go to the data block that begins with MOLLY.

Legend

Note that data records are never chained together in Enscribe key-sequenced files. Instead, the

tree structure is dynamically rebalanced to ensure that all records in the file can be accessed with

the same number of FILE_READ64_/READ operations, that number being the number of levels of

indexes plus one for the data block.

Unique Features of EKS Files

Enhanced Key-Sequenced Files (EKS) have these unique features:

• The primary partition does not store user data but rather stores file meta data

• When creating an EKS file, the partial key of the first secondary partition must be explicitly

set to all zeros.

70 Key-Sequenced Files