Filter
Top Products
SSL Linkage methods
U929-J-Z125-9-76 157
Dokuschablonen 19x24 Version 7.3us für FrameMaker V7.x vom 14.02.2007 © cognitas GmbH 2001-2007
24. Oktober 2007 Stand 08:58.55 Pfad: G:\vogt\fsc\uds\Manuale\en\udsent_e\udsent.k05
5.3.3 Preventing redundancy in SEARCH key tables
TYPE IS DATABASE-KEY-LIST
In the description of the logical data structure drawn up using the schema DDL, the user
decides if UDS/SQL is to set up a SEARCH key table (SEARCH KEY IS... USING INDEX).
The form of this table can be influenced by means of the TYPE clause.
The UDS/SQL standard form of SEARCH key table is that of a multi-level pointer array: It
consists of one line for every record of the associated record type or for each member
record of the associated set occurrence. This line contains the key value of the record and
the pointer to the record.
(This standard form corresponds to the specification: TYPE IS REPEATED-KEY.)
Different records can often have the same key values (DUPLICATES ARE ALLOWED). In
such cases, TYPE IS DATABASE-KEY-LIST can be specified; this ensures that a key value
with duplicates is stored only once.
This type of SEARCH key table is referred to as a duplicates table. Duplicates tables are
useful
– in the case of long keys,
– if several key values have between 5 and 2000 duplicates,
– when processing large numbers of records (see the "Application Programming"
manual, FIND 7).
A third overflow page must be generated by UDS/SQL in the following cases:
– if there are more than approximately 2000 duplicates in a 2-Kbyte or 4-Kbyte database
– if there are more than approximately 4000 duplicates in an 8-Kbyte database.
This has an adverse effect on the access behavior (one additional access per overflow
page), which should be weighed against the saving in space achieved.
Duplicates tables are set up by UDS/SQL on the lowest level only. Higher table levels are
comparable to those of a standard SEARCH key table.