Synchronization in a parallel-accessed data base

Abstract

The following problem is considered: Given a data base which can be manipulated simultaneously by more than one process, what are the rules for synchronization which will maximize the amount of parallel activity allowed. It is assumed that the data base can be represented as a graph. An example of such a data base is a hierarchy of directories for an on-line file system. Methods for synchronization of processes are examined; their validity is discussed and their performance compared.