Although the amount of available data is growing, decision making remains a challenging task due to the number of factors and their interdepencies. While such problems are well-known in business life we can find similar questions while we try to understand and control the IT systems. Moreover the previous problem could be harder, because often the configuration or the system workload characteristics may vary over time. Modern monitoring methods typically gather a huge amount of raw data. The information behind this data can be complex and hard to interpret for an operator. Modern methods of data analysis and decision support used in this work can help in decision making, while this process can partly be automated as well. Such decisions can help in identification of necessary system improvements or determine monitoring points. The models created for complex systems are also an important as these should be portable so that reuse in other configurations can be supported. In this work I present a method to apply modern technologies of decision support and data analysis in engineering decisions. The input data is gathered from the simulations and the performance measurements of the systems. I also show how we can use the previously mentioned methods for making the models. I show how we can use the decision support methods while taking account the informations gathered from the system models and expert knowledge. I also evaluate these methods and analyse their effectiveness on a real case study. I also show how the results of such methods can be used in synthesizing performance models.