Concurrency in linear hashing

Concurrent access to complex shared data structures, particularly structures useful as database indices, has long been of interest in the database community. In dynamic databases, tree structures such as B-trees have been used as indices because of their ability to handle growth; whereas hashing has been used for fast access in relatively static databases. Recently, a number of techniques for dynamic hashing have appeared. They address the major deficiency of traditional hashing when applied to databases that experience significant change in the amount of data being stored. This paper presents a solution that allows concurrency in one of these dynamic hashing data structures, namely linear hash files. The solution is based on locking protocols and minor modifications in the data structures.