We are interested in taking into account ambiguity in the analysis phase of data warehouses, using fuzzy logic. We want to offer decision makers the possibility of using natural language in this phase. We created in a previous work the Baccalaureate fuzzy data warehouse which we were able to query with seven natural language terms to which we created seven membership functions. In this work, we present a fuzzy data warehouse for server failures that we created and for which we used the same terms to which we associated seven membership functions too. And, we carried out a comparison at the end of which we concluded that the definition of the values of the membership function differs according to the context of analysis. Our solution is extensible and can be enriched with new natural terms language. The next step is to design a conversational interface that enables a natural language conversation between the decision maker and the fuzzy data warehouse.