Power transformers’ failures carry great costs to electric companies. To diminish this problem in four working 40 MVA transformers, the authors have implemented the measurement system of a failure prediction tool, which is the basis of a predictive maintenance infrastructure. The prediction models obtain their inputs from sensors, whose values must be conditioned, sampled and filtered before feeding the forecasting algorithms. Applying Data Warehouse tech- niques, the models have been provided with an abstraction of sensors the authors have called Virtual Cards. By means of these virtual devices, models have access to clean data, both fresh and historic, from the set of sensors they need. Besides, several characteristics of the data flow coming from the Virtual Cards, such as the sample rate or the set of sensors itself, can be dynamically reconfigured. A replication scheme was implemented to allow the distribution of demanding processing tasks and the remote management of the prediction applications.