Evaluating the quality of service of a data warehousing system

Abstract

Data Warehousing Systems and Business Intelligence applications in general are some of the top priorities for many companies in the world. During their existence many of them have been spending large amounts of money in specific decision support architectures and related infrastructures in order to speed-up and scale-up their decision making services, and improve performance of their decision agents. However, in spite of being optimized systems – as expected – these expensive architectures not always provide fast and correct answers for ad-hoc queries, reporting services, or even fast processing times for multidimensional structures. A lot of reasons could explain such undesirable and not justified performance, ranging from complex cases of bad index structuring and join paths to simple overlapping of processes, running at the same time over the same analytical processor. As we know, such things are not simple to identify and solve in a convenient and practical way in effective decision-making scenarios. In order to sustain the quality of service of data warehousing systems we could use profiling and prediction strategies to identify and observe relevant system performance measures that provide us the means to establish better operational constraints. Based on this, we have designed and implemented a software tool especially oriented to predict and regulate the quality of service of a data warehousing systems, providing to their administrators the basis to establish more effective decision-making support services and the means to attenuate quality of service degradation.