Abstract:
A workload characterization system characterizes an exemplary business intelligence workload for use in sizing a hardware configuration required by a new database system running a similar business intelligence workload. The workload characterization system uses performance-oriented measurements to characterize an exemplary workload in terms of resource usage and performance metrics. The workload characterization system applies unsupervised data mining techniques to group individual business intelligence queries into general classes of queries based on system resource usage, providing insight into the resource demands of queries typical of a business intelligence workload. The general classes of queries are used to define an anticipated workload for a planned database system and to help identify the hardware required for the planned database system.