Semantic-based concurrency control protocol and distributed software agent for real-time database systems.

摘要

Real-Time transactions have both logical and temporal consistency constraints. In addition to maintain the database integrity and consistency as in conventional DBMS's, a real-time database system (RTDBS) must be able to handle timing requirements for transactions and for data objects. As a result, the transaction scheduling policy and concurrency control algorithm in such systems play crucial roles on their performance. In this dissertation, we propose a semantic-based concurrency control approach which can be used to improve the performance, especially the external consistency requirement, of an RTDBS. It employs simple, flexible and powerful enabling techniques that can improve the desired system performance for different RTDBS applications without incurring unacceptable operational overheads. In particular, we study the impact on two mission-critical systems, namely air traffic control systems and radar warning receiver systems.; We also study a software agent model to integrate RTDBS with distributed computing. A mobile agent carries out specific execution code on client's or server's behalf. It can enhance the system efficiency if appropriate scheduling algorithm is employed, which is the key to comply with the temporal constraints of real-time transactions so that better or guaranteed quality of service can be achieved. We study different scheduling paradigms on both centralized and distributed environments. In addition, we investigate the implementation issues of a real-time Web-based auction prototype system.