|
There are several techniques that can help
to deliver a successful data warehouse. These fall into three
broad groups:
Modelling
techniques
Business process modelling can be used to help understand
the business and identify candidate data marts. It is also
useful for communicating the way data flows from one stage
to the next on its way from source system to user (see components
diagram).
Entity
relationship modelling will be needed by the data
staging team to understand how data is organised in each of
the source systems. This is a standard systems analysis technique
used to build OLTP
systems.
Multi-dimensional data modelling (or dimensional
modelling, as it has come to be known) will often
be the primary specialist technique used to design the central
warehouse and associated data marts.
The key steps in designing a dimensional data warehouse are:
- Identifying the major processes and hence the required
fact tables;
- Deciding the granularity
and dimensions
of each fact table;
- Defining the measures
needed for reporting and analysis, for each fact table,
including derived measures and full descriptions;
- Agreeing the attributes, levels and hierarchy
for each dimension, including all labels and full descriptions;
- Deciding how to track slowly
changing dimensions;
- Selecting the aggregates
that will be provided as physical data stores;
- Deciding how much historic data to keep, and how often
the warehouse needs to be updated.
One of the most difficult aspects of this process is often
achieving consensus across the organisation on the standardised
definitions and terminology to be used.
Kimball’s
books include excellent introductory and advanced material
on this topic, along with numerous practical tips and examples.
The most important issues are highlighted in the section on
design issues.
Project
management techniques
Good project management is essential for any data warehousing
initiative. Perhaps the most crucial factors for success are:
- establishing an effective partnership between the business
and technical contributors;
- securing and maintaining support from senior management.
The books by Adelman
and Kimball
both discuss the planning and management of data warehousing
projects in some depth. Other useful ideas and approaches
can be found in several best practice methodologies, including
the following:
PRINCE 2
The CCTA’s PRINCE
2 methodology has particularly helpful approaches to
establishing the business case, defining project
team roles and managing risk.
Programme Management
Recent thinking on programme management is also pertinent,
especially where the warehouse is an enabler for a wider
strategic objective e.g. improving customer
relationship management. The most useful ideas concern
stakeholder management and benefit realisation.
For more information see the CCTA guide to "Managing
Successful Programmes".
Dynamic Systems Development Method
DSDM is a recognised methodology for rapid application
development, and is used by a number of consultancies operating
in the data warehousing and business intelligence arena.
It places a strong emphasis on user involvement, and uses
ideas such as time-boxing
and prototyping, both of which can be very effective in
a data warehouse environment.
For more information, refer to the DSDM
Consortium.
Communication
skills
Perhaps the most important step in building a data warehouse
is to understand what managers want to do with the data. This
requires good interviewing skills.
Another challenge is the resolution of conflicting objectives,
priorities and terminology. This requires good facilitation
skills.
|
