This process, called ETL (Extraction, Translation/Transformation, Loading),
is executed by a software layer located between data sources and a DW (Kimball &
Caserta, 2004; Simitsis, Vassiliadis, Terrovitis, & Skiadopoulos, 2005). The software
includes: monitors that are responsible for detecting and reading changes in data
sources, wrappers that are responsible for transforming a source data model into a
common DW model as well as an integrator that is responsible for integrating data,
cleansing them and loading them into a DW (Widom, 1995).
x
The initial loading into an empty DW reads all data of interest from EDSs and stores
them in a DW. As the data sources are operational systems that are used everyday,
their content changes frequently. As a consequence, the content of a DW becomes
obsolete and has to be refreshed. A data warehouse is often implemented as the
collection of materialized views, thus the problem of a DW refreshing transforms
to the problem of maintaining and refreshing materialized views. This problem has
been extensively investigated by the research community (Gupta & Mumick, 1999;
Roussopoulos, 1998) and has resulted in multiple algorithms for refreshing materialized
views. Typically, refreshing a materialized view is a costly task. In order to
optimize it, multiple incremental refreshing techniques have been proposed.
Pages:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25