From Conventional to Multiversion Data Warehouse - Evolving Application Domains of Data Warehousing and Mining - page 41

Chapter 3
From Conventional to
Multiversion Data Warehouse:
Practical Issues

Khurram Shahzad
Royal Institute of Technology (KTH)/Stockholm University (SU), Sweden

ABStrAct

The Data warehouse is not an autonomous data store, because it depends upon its operational source(s)
for data population. Due to changes in real-world scenarios, operational sources may evolve, but the
conventional data warehouse is not developed to handle the modifications in evolved operational sources.
Therefore, instance and schema changes in operational sources cannot be adapted in the conventional
data warehouse without loss of information. Multiversion data warehouses are proposed as an alterna-
tive to handle these problems of evolution. In this chapter we discuss and illustrate how versioning is
implemented and how it can be used in practical data warehouse lifecycle. It is designed as a tutorial
for users to collect and understand the concepts behind a versioning solution. Therefore, the purpose
of this chapter is to collect and integrate the concepts, issues and solutions of multiversion data ware-
houses in a tutorial-like approach, to provide a unified source for users that need to understand version
functionality and mechanisms.

IntroductIon Data is extracted from OLTP(s), transformed

and loaded in data warehouse after removing in-
consistencies. Therefore, the data warehouse is an
integrated and materialized view of data, which is
optimized to support decision making (Chaudhuri,
1997). The data warehouse works as a data source
for various types of applications e.g. analytical
processing, decision making OLAP, data mining
tools, dashboards etc.

Online transaction process systems (OLTP) are used
to meet day-to-day requirements of an enterprise.
But OLTPs' are unable to meet decision support
requirements of an enterprise, because a) their sche-
mas are not optimized to support decision-support
queries and b) they are not made to support decision
making (Paulraj, 2001; Kimball, 2002).