Reai-Time Index Concurrency Control

Research on real-time database systems (RTDBS) has been underway for close to a decade now. So far, this research has focused mainly o11 identifying the appropriate choice, with respect to meeting the real-time goals, for the various database system policies such as priority assignment, transaction concurrency control, memory management, etc. (see [16] for a recent survey). However, the design and evaluation of concurrency control algorithms for indexes, which are an integral part of database systems, has received virtually no attention in the real-time environment. In our view, this lacuna is rather surprising since it appears reasonable to expect that indexes would be one of tbe primary mechanisms by which a well-designed RTDBS would try to meet transaction timing constraints. In this situation, using an "offthe-shelf" index concurrency control algorithm that has not been specifically designed for the real-time environment may lead to many transactions missing their deadlines. To address the above issue, we have initiated a project called RICC (Real-Time Index Concurrency Control), whose goals are to identify the basic forces that determine real-time index performance and to develop index concurrency control algorithms that are tuned to the real-time environment. This project is being jointly conducted at the Indian Insitute of Science (Bangalore) and at the Indian Institute of Technology (Bombay) in India. In this article, we present an overview of the results obtained so far and identify future research avenues.