© 1976 by British Computer Society
| ||||||||||||||||||||||||||||||||||||||||||||||||||
The problem of lock by value in large data bases
Universitat Karlsruhe, Institut fur Angewandte Informatik und Formale Beschreibungsverfahren, Karlsruhe, Germany
In parallel-accessed database systems a process must be able to lock a subset of the data. In current systems and proposals the only way for locking a set of data is either to lock explicitly each element in the set of to lock a large predefined set, such as a whole file. Though the necessity of a more flexible locking feature is recognised there are no results as to conceptual foundation and implementation. In this paper a concept for lock by value is developed which allows a process to lock a subset of the database just by specifying a condition describing this set. Thought the concept looks rather simple some unpleasant problems have to be discussed. A locking system supporting current lock features at the record-level as well as lock by value is proposed.
Received June 1975.
* Universität Karlsruhe, Institut für Angewandte Informatik und Formale Beschreibungsverfahren, D 75 Karlsruhe, Postfach 6380, Germany