Why Column-Wise Ntuples ?

• Column-Wise Ntuples base analysis program reads only data necessary in a given instance of the analysis program, hence the I/O is reduced to absolute minimum.
• Packing facilities provided allow to achieve a major reduction of disk space (of the order of factor three over the standard microDST), hence further reducing the the I/O load of the analysis jobs, and permitting more efficient use of the available disk space.

Why PAW ?

• A lot of the analysis work is done on all kinds of N-tuples. Most of it can be relatively easily transported to the Challenge.
• Use of PAW interface will make the transition to UNIX operating system almost invisible (for all these, whose world does not extend beyond PAW). In any event, it will permit to continue the efficient physics analysis during the 'transition-to-UNIX' phase.

Why PIAF ?

• Client-server form of running PAW will allow use of local desktops (be it VAX, Alpha, Indy or PC) to run the analysis job, and import the results to the favoured presentation package.
• Minimal network traffic between the PIAF server and the PAW clients will enable to run the analysis jobs equally efficiently at any remote Institution as at Fermilab.