The rebuild process essentially is a total re-creation of the data warehouse. One of the major impacts of such an approach is the end user layer, or rather the effect on the end-user tools and saved queries and reports that are currently in use. The redesign or redeployment of this interface to the end users of the data warehouse may be too large a task to undertake. The problem can be circumvented to some degree through the use of views to make the new data warehouse environment look the same as the previous. But it is this impact and the subsequent re-testing process that must be considered when deciding to undertake a rebuild.
The advantages of a rebuild are the seamless integration of future analysis areas into the data warehouse and the single point of management that is provided. The major steps in the rebuild process will depend very much on the environment being replaced. As a guideline, the following steps may be worth considering.

The rebuild process

  1. Load a copy of the WhereScape metadata repository into an otherwise empty test environment. Refer to the RED Installation Guide for instructions on how to load a metadata repository.
  2. Ensure there are no public synonyms that point to existing table names if the rebuild process is to use the same names as some or all of the existing tables.

Working within the WhereScape RED tool proceed to:

  1. Create connections to the new test data warehouse and to all the source systems.
  2. Using the source system knowledge from the existing data warehouse, create the appropriate load tables in the data warehouse based on the existing extract or load methodology.
  3. Build up the dimension, stage, and fact tables using the same column and table names where possible.
  4. Examine the existing procedures or update methodology and include this in the generated stored procedures.
  5. Test the new environment.
  6. Work out a plan to convert the existing data into the new data warehouse. Where possible it is best to keep existing key values and re-assign sequences to match these existing key values where appropriate.
  7. Convert and test the old data warehouse data in the new environment.
  8. Redeploy the end-user tool access.
  • No labels