A method for relational database scheme design with the aid of a digital puter for a database having attributes A.sub.i, i=1 to n and relational schemes R.sub.j, j=1 to m. Each relational scheme R.sub.j is a non-empty subset of the attributes A.sub.i. The method detects any scheme that is non-acyclic in a simple manner that is easily adapted to a digital computer environment. The resulting relational database scheme design is thereby prevented from being non-acyclic.
A database operation control system includes a database checker, a database saver, a database diagnoser, a logical image creator, a database creator, and a database restorer. The database checker checks an access time taken for retrieval processing, and records an access time increase in an access time record file. The database saver saves data from the database to a save file. The database diagnoser compares the access time increase with a reorganization start increase as a reference value obtained from the access time record file to diagnose whether recognition is necessary, and informs the database saver of the necessity of the execution of reorganization processing of the database, if it is determined that reorganization is required. The logical image creator decomposes data of the save file into logical units to create logical images, and stores them in a logical image file. The database creator creates a restoration file. The database restorer restores data from the restoration file to the database. Reorganization processing is automatically performed in accordance with the access time increase obtained by the database checker.
A method for implementing an acyclic directed graph structure using a computer relational database system, and a data structure corresponding thereto. An advantageous relational database structure representing the acyclic diagraph can be obtained by creating three table structures storing relationships between nodes in the diagraph and other properties. A node table stores relationships between each node in the diagraph and properties assigned to the nodes. An edge table stores indications of nodes which are directly linked and the direction of the link. A path table stores indications of nodes which are directly or indirectly linked and the direction of those paths. Procedures for maintaining the database structure in a consistent manner are also described.
A method, apparatus, and article of manufacture for optimizing a star join operation in relational database management systems (RDBMS). A cross-product is generated from a plurality of dimension tables referenced by the star join. The join columns of the cross-product are then hashed to create a hash-row value. Using the hash-row value, a Star Map is probed to determine whether a record exists in a fact table that corresponds to the cross-product, wherein a first portion of the hash-row value is used to select a row of the Star Map and a second portion of the hash-row value is used to select a column of the selected row of the Star Map. The fact table is accessed to perform a merge join with the cross-product when the selected column of the selected row of the Star Map indicates that the record exists in the fact table.