Data Warehouse Architecture - Evolving Application Domains of Data Warehousing and Mining - page 13

quality must be considered multiple times during
the whole enterprise data lifecycle. In the data
warehouse architecture, quality of data after each
“loading data” process should also be checked. The
definition of data quality rules must be consistent
with the enterprise's data quality discipline. If a
SOA method is applied to the data staging area,
the data cleansing and standardization operations
in the data staging area are either services provided
by other parties through the enterprise service bus
or providing services to other parties through the
enterprise service bus.
Second, the integrated data model at the enter-
prise data warehouse must be reconciled with the
model of master data. One challenge of managing
the consistency between the two models lies in that
the data model in MDM is normally concerned
with the current data while the data model in the
enterprise data warehouse covers a longer his-
tory of data. Connected to the data models, there
should be a single and consistent data governance
model over both the master data and data in the
data warehouse.
Third, the typical workflows around data ware-
house architecture are normally started from an
activity triggered from operational system, such
as a data delivery or change of metadata. And
these workflows normally finish by an activity at
the data access tools side, such as a report that is
generated or an analytical result is produced. When
MDM is considered, the workflows have to be in-
tegrated with activities at the operational systems.
Extending data warehouse architecture into SOA
architecture is capable of facilitating this integra-
tion process. In service-oriented data warehouse
architecture, different data warehouse services
can be easily assembled into workflows.

have to be processed and managed in a timely man-
ner to fulfill vast amount of business requirements.
The rapid-growing hardware improvement, new
internet infrastructures and emerging computing
frameworks are facilitating the future development
of data warehouses while posing fresh challenges
to the architecture design of data warehouses.
This chapter is dedicated to review existing in-
dustry practices on data warehouse architectures
and introduce new challenges and trends to the
future of architecture design patterns for data
warehouses. Compared to the classical theory of
software and information system architecture,
data warehouse architecture is more dependent
on the different properties and aspects of its key
asset, the data. Different architectural patterns in
normal software system design have to be tailored
to meet the demands of efficient management of
data. As a conclusion of the chapter, it becomes
an urgent need to both the information system
architecture and database communities that a
data-centric architecture discipline has to be settled
for further development of data warehouse and
database architectures.

referenceS

Adiba, M. E., & Lindsay, B. G. (1980). Database
Snapshots. In Proc. Of VLDB (pp. 86-91).

Agrawal, R., Gupta, A., & Sarawagi, S. (1997).
Modeling Multidimensional Databases. In Proc.
ICDE (pp. 232-243).

Ballou, D. P., & Tayi, G. K. (1999). Enhancing
Data Quality in Data Warehouse Environments.
[CACM]. Communications of the ACM , 42 (1),
73-78. doi:10.1145/291469.291471

Bell, M. (2008). Introduction to Service-Oriented
Modeling. In Service-Oriented Modeling: Service
Analysis, Design, and Architecture . Hoboken, NJ:
Wiley & Sons Inc.

concluSIon

The research and development of data warehous-
ing technologies have evolved into the era that
complex, high volume, and various types of data