Text
Search: General-Purpose DBMS vs. Specialty
Search Engines
While there are exceptions to any rule, we generally
favor building text search applications on top of general-purpose DBMS such as
Oracle or DB2 -- rather than specialty search engines.
First of all, manageability
concerns generally favor the general-purpose DBMS. Every
large enterprise has tools and procedures to manage DBMS, back up data, and so on. Generally, it is safer and easier to extend this
investment to some new applications and tables than it is to bring in a new specialty text
database engine.
Second, security
is more flexible within a general-purpose DBMS. Some
of the most important applications of Web-style search are for the dissemination of
technical or customer information, either internally or to an extranet. Such information is in many cases proprietary and
sensitive. And it may be hard to do a great
job of sorting out entitlements and permissions without a lot of hooks into human
resources and/or customer information systems. Indeed,
many text search applications both intranet and internet would be much
enhanced in utility if sensitive and not-so-sensitive information could be searched
side-by-side. And the best way to make that
happen is with DBMS-based security.
Security even aside, there simply are a lot of
application areas in which text search is or should be tightly integrated with traditional
RDBMS-based application functionality. Examples
include:
Resume
management
Knowledge
management (specifically, the part of KM which works, namely finding internal experts in
your enterprise)
Search-based
navigation (think of Amazon.com)
CRM
of all kinds (comment fields on call reports, maintenance reports, etc.)
Product
lifecycle management
Meanwhile, the traditional reasons for preferring a
specialty text engine to a general-purpose DBMS are largely obsolete. Performance
is usually a non-issue. And while specialty
products are rich with text-specific features,
these are usually available in general-purpose DBMS as well.
For more discussion of these issues, please see our
section on text search.
For more information, please contact Curt Monash or Linda Barlow.
To reach Monash
Information Services by phone, please call 978-266-1815.
Copyright 1996-2003, Monash Information Services. All
rights reserved.
Updated: 05/11/04 |