A data warehouse is a business database where historical and current business data from disparate source systems has been consolidated into a central place.  It is the foundation stone upon which a comprehensive BI solution is built as it becomes the source for all reporting and analysis.  As such it improves consistency of reporting across the business, perpetuation "one version of the truth".

As the data warehouse underpins the rest of your BI system, it is imperative that it is designed, loaded and maintained properly. The approach to designing a data warehouse is very different from a transactional system and requires very different skills and techniques. 

Why build a data warehouse?

The key objective in building a data warehouse is to create a “single version of the truth” whereby numbers are reported consistently across the enterprise.  This avoids scenarios where different figures are reported at important meetings for apparently the same business measure.

Building on what you have

If you currently have a ‘reporting system’ where data is extracted from your main systems and prepared for reporting, then you already have a data warehouse of sorts.  It may not be called a data warehouse and may not have a coherent architecture, but it may well achieve some of the goals of a data warehouse.

We will review any existing system you have in place and recommend the best approach.  This may be anything from extending functionality to re-architecting the system.  We can also facilitate knowledge transfer so your existing team can become familiar with best practice in data warehouse design.

Data warehouse architecture

There is no single data warehouse architecture to which all systems must comply.  Instead, there is a set of common functions that every data warehouse needs.  How much time and effort is dedicated to each function will depend on the scale of your implementation and your regulatory and audit requirements.  For example, there would be no need to build sophisticated data cleaning utilties if your corporate data is already clean!  On the other hand you may well have strict audit requirements that mean any data change is recorded and validated independently.  Getting the data right is usually where 70% of the project effort is expended!

Components of a data warehouse

Business rules dictionary provides a name and description of each measure and how it is obtained or calculated.  A measure being a numerical fact about your business that you wish to monitor (e.g. number of units sold).  The business rules dictionary will also define business metadata and key business terminology.