| Summary: | Data warehouses have been proven useful in making strategic business decisions by
enabling storage of large amount of data, possibly from heterogeneous and distributed
sources, into a unified system. The data kept in the data warehouse is historic mainly
because of the batch process that is used to update the data. However, several business intelligence applications now require access to real time data, in addition to
historic data. This leads to the need for a real time data warehouse. But questions arise as to how we can achieve real time data warehousing and if the current
approaches address the requirements of a real time data warehouse.
We look at real time systems and real time databases, examine their nature and
their issues, and synthesize the requirements that are applicable to real time data
warehousing. By comparing and combining, we have come up with a list of requirements a real time data warehouse must satisfy. A real time data warehouse requires
quality of service (QoS), quality of data (QoD), and some application-based requirements. For each requirement, we discuss what component of the data warehouse can be used to address the requirement. As we discuss each component, we also present
an overview of the state of the art. With the requirements, current approaches, and
limitations known, we highlight what areas need further investigation to make real
time data warehousing a reality. Our study targets researchers who would like to
contribute to the advancement of real time data warehousing, and data warehouse
administrators of business intelligence systems that need to be made real time.
The second half of the author's work focuses on conceptual modeling for real-time
data warehouses. A lot of work has been presented in the area of conceptual modeling
for designing data warehouses. However, in spite of the advent of real-time data warehouses, not much work can be seen with regards to modeling of these systems.
In this context, we propose a conceptual modeling approach RealMER (based on
Entity-Relationship), to address the peculiarities associated with this kind of systems.
We firstly propose a set of criteria for classification of good conceptual models for
real-time data warehouses. Then, we define new graphical elements to cover the
real-time paradigm and highlight the advantage capturing such information has on
the downstream processes of the warehouse. Through a case study and comparative
analysis, we prove the effectiveness of our technique over those of our peers.
|
|---|