The determination of efficient record segmentations and blocking factors for shared data files

It is generally believed that 80 percent of all retrieval from a commercial database is directed at only 20 percent of the stored data items. By partitioning data items into primary and secondary record segments, storing them in physically separate files, and judiciously allocating available buffer space to the two files, it is possible to significantly reduce the average cost of information retrieval from a shared database. An analytic model, based upon knowledge of data item lengths, data access costs, and user retrieval patterns, is developed to assist an analyst with this assignment problem. A computationally tractable design algorithm is presented and results of its application are described.