Data management and concurrency control in broadcast based asymmetric environments

Abstract

Tens of millions of users have personal handheld devices with several network interfaces built-in, and the number of users and of network interfaces included are only increasing. This growth suggests a need for new methods to disseminate data to multiple clients, and cyclic broadcast is one approach. We do a survey on the various data management protocols that describe how to broadcast the data, and the concurrency control protocols that make sure all access to the database is consistent. The various data management and concurrency control techniques deals with the restrictions in asymmetric broadcast environments and improves the performance. As part of the survey, we made a simulation platform and implemented several data management techniques and four concurrency protocols, BCC-TI, FBOCC, PVTO and STUBcast. We also implemented a technique we call “partial restart” into FBOCC. Simulation results shows nearly a 30% decrease in transaction execution time when the transaction length is 6. The survey and the simulations helped us to make a qualitative characterization framework, which categorize the protocols based on the environments they apply to and qualities they possess. We offer the source code to the simulation models and the protocols we implemented which can be used as a base to implement new protocols.