IBM Support

IBM DB2 Text Search Version 10.5 : new features and enhancements



This document describes the new features and enhancements in DB2 Text Search V10.5.


DB2 Text Search enables an IBM DB2 for Linux, UNIX, and Windows database user to create applications with full text-search capabilities by embedding full text-search clauses in SQL and XQuery statements. Its new generation text search component from IBM.
The functionality of Text Search is enhanced/extended in V10.5.

Command line tools:

DB2 Text Search 10.5 provides enhanced version of command line tools and the details are explained below:

Config Tool
Enhanced configuration capabilities for example, ability to define system level defaults which would apply to all index collections, ability to change all the configuration parameters via the configTool etc are added newly.

The configuration changes for V10.5 set by the user using the configTool will be updated to sqllib/db2tss/config/configuration.xml instead of the old file sqllib/db2tss/config/config.xml.

For more information on each configuration parameter, see the information center.

Admin Tool
The trace parameter is deprecated in this release, use -logLevel instead to enable/disable the trace.

For more information on this tool, see the information center.

Synonym tool
Now, we can also specify the language for which synonyms are defined using -language <2-character language code> parameter.

For more information on this tool, see the information center.

Search enhancements:

Special characters

You cannot search for an exact match on two consecutive, identical special characters. Queries of this type return documents that contain only one of the special characters.

For more information on special characters, see the information center.

Structural full-text search in XML documents

The following search features are supported by XML search:
Boolean operators (basic search)
exact match
fuzzy search
proximity search
stop words
wildcard characters

For more information and examples on xml search, see the information center.

Trouble shooting Text Search:

Logging and tracing for the DB2 Text Search server

There are changes in the steps to collect the data required to trouble shoot text search related problems, for more details please refer to the link:

For more information on logging and tracing, see the information center.

Monitoring queues for DB2 Text Search index updates

You can gather more monitoring information now with this release and this can be used to tune the DB2 Text Search server configuration, if you experience indexing performance issues.

For more information on monitoring queues, see the information center.

Control update index processing:

To control the update behavior several options are available now, from specifying a commit-batch to secure completed work, to parameters that determine the number of maximum parallel update processes, maximum collection numbers, etc. With this release, the administrators can control the time-window that an index update may use and thus avoid load on the system during peak time.

The following capabilities are enabled with this release:
  • Ability to specify the commit cycle in hours to simplify the specification of commit-batches
  • Provide an option to gracefully end the index update and continue processing the next time the update starts by enabling the use of a maximum time window for updates to prevent that index processing continues to run during peak workload times
  • Simplify schedule handling by combining create and initial update
  • Allow control of which data to process in the execution of the initial update
  • Enable an option to specify a custom logtype to control whether triggers are created and how the auxiliary logtable is defined
Commit batches

Text index update processing already provides a feature to specify the commitsize via the updateautocommit argument. To provide further control, additional settings are available now to determine whether the commitsize should be treated as rows or hours, and how many batches to process.
Each commitcyle comes with a significant overhead, and should only be used for very long-running index updates.

For more information on commit batches, see the information center

New index configuration options

A new option 'INITIALMODE' is introduced in this release, with this option, you can run the initial update immediately when the index is created. Also, you can defer it to the first update operation or skip it altogether.
Another option LOGTYPE' is introduced to set a custom logtype so that the text index administrator can decide whether triggers should be added to populate the primary log table.

For more information on these new index configuration options, see the information center

Set command locks:

It is possible to set the command lock manually, for example, to restrict access when an administrative operation is applied on the collection level.
For more information on command locks, see the information center.

Some of the other capabilities include the following:
  • Enhanced error handling, specifically query syntax errors
  • Enhanced indexing and optimization changes which would speed up indexing and optimization, and reduce impact of indexing/optimization on search.
  • Stronger support for multilingual collections (provide more accurate results across languages)
  • Support for embedded documents, archive files, such as ZIP, RAR and TAR files by extracting, concatenating, and indexing their content.

Document information

More support for: DB2 for Linux, UNIX and Windows
Extenders - Text

Software version: 10.5

Operating system(s): AIX, HP-UX, Linux, Solaris, Windows

Software edition: Advanced Enterprise Server, Enterprise Server, Workgroup Server

Reference #: 1643180

Modified date: 20 August 2014

Translate this page: