Abstract

Data warehouses are databases designed to help organizations in the decision-taking process. Data 
warehouses combine data from several other database and transaction processing systems to make 
use of the data as a whole. So that the data warehouse represent the organization’s reality they need 
to be updated periodically. That updating process may required quite a few resources and 
sometimes shut down the data warehouse so that analyst can not access it. Analysts require that data 
warehouse be up all the time so that data warehouse maintenance process is a critical point of the 
system. For that reason research about efficient strategies to improve data warehouse maintenance 
has recived special attention since this technology appeared.

Data warehouse maintenance is a three-phase process: Draw out of data from their sources, data 
transformation, and data warehouse updating. 

This research work has to do with the data transformation phase and mainly with the updating one. 
For the transforming phase a system has been developed to perform data moderate cleaning 
activities, data format integration and data semantic integration. As to the updating phase two 
algorithms were defined and implemented to perform data warehouse updating on both incremental 
and line approaches. The use of those algorithms does not require that the data warehouse be down 
during the maintenance phase.
The algorithms used in the data warehouse updating phase are based on a multi-version strategy that 
let having an unlimited number of the updated data so that users may access the same data version 
while the warehouse is being updated. Those algorithms improve other existing literature 
algorithms and let data warehouse maintenance in either real time or batch. 

Benefits of this research are quite significant due to it is improving the use data warehouse 
technology which is under huge growing fashion in all corporate processes mainly in the supply 
chain management process.