Data mining with PIAF

Using the Parallel Interactive Analysis Facility, PIAF, users can query in real-time multi-gigabyte databases. Interactive response times are obtained by executing queries in parallel on a cluster of high speed workstations. For computer intensive queries an almost linear speedup is achieved using five workstations in parallel. In this paper the authors describe their Ntuple database model, the PAW system used to query, analyze and visualize the data as well as the hardware and software aspects of the PIAF system.