Integrity checking in data base systems

Abstract

There is the problem, how to control data base integrity. For many types of integrity assertions it is an unsolved problem, how to prove them in justifiable time. This paper deals with methods for improving the testing of semantic integrity assertions. First a classification of integrity assertions is given and a proposal for the implementation of a subsystem for controlling semantic integrity is described. Special algorithms are presented for testing delayed integrity assertions after the end of transactions, for testing integrity assertions, which contain functions, and for testing integrity assertions, which define relationships between values of attributes of different tuples. Using these algorithms in many cases the integrity test can be performed without any access to secondary storage, so that the cost for the tests becomes justifiable.