Semantic Concurrency Control on Continuously Evolving OODBMS Using Access Control Lists

Abstract

Object oriented databases (OODBMS) are widely used for applications which require support of complex relationships on data. Transactions access the database for data or schema simultaneously. The runtime transactions read/modify the data. The design time transactions read/modify the schema. In a continuously evolving domain, more number of design time transactions is executed along with runtime transactions. Parallel execution of runtime and design time transactions affects the consistency of the database. Hence a concurrency control technique is needed to preserve the database consistency. Several concurrency control techniques using multi-granular lock models have been proposed in the literature. Multi-granular lock model provides better concurrency and is simple to implement. But high concurrency is provided to dynamic databases at the cost of heavy maintenance overhead. In this paper, a concurrency control scheme is proposed using access control lists to provide better concurrency, without the maintenance overhead. The performance of proposed scheme is compared with existing work. It is found that the proposed scheme gives better response time than the existing work and also eliminates the overhead of maintenance of access vectors.