An integrated Architecture for Business Intelligence support from Application down to Storage

Recent developments both in the business and the technological domain have led to a significant increase in demand for Business Intelligence (BI) infrastructures that can handle huge amounts of data in small time frames. BI applications are increasingly used by large user bases on all management levels; support tasks spanning the complete value chain are based on transactional data and are directly coupled with operational systems in ”closed loop” approaches.<br><br>To effectively handle the resulting data volume turns out to be an extremely challenging task which encompasses a variety of issues on different levels. We propose an integrated multi layer tool for monitoring, benchmarking, analyzing, and optimizing the performance of such BI infrastructures.<br><br>Inside this paper we give a coarse outline of the tool’s architecture and demonstrate the value of distinct measurement points at operating system layer. For that purpose we introduce a kernel based benchmark environment and present first measurement results. The gathered data clearly indicates that a meaningful analysis of performance benchmarks without kernel trace points is of limited value - which shows the necessity to consider a separate component within the tool’s architecture.