A Case for Real-Time Client-Server Databases

Abstract

The support of applications in large-scale financial environments [20], wireless directory services, and Personal Communication Systems (PCS) [18], requires systems that efficiently handle high volumes of data and demonstrate real-time processing features. Two research areas that could assist in the realization of the above goal, Real-Time Systems (RTSs) and Client—Server Databases (CSDBSs), have individually attracted a lot of attention during the last few years. In conventional soft Real—Time Systems, tasks are submitted for execution based on deadlines imposed by the applications. RTSs offer improved response times by utilizing elegant scheduling techniques and exploiting a-priori knowledge of the nature of the tasks [1, 10, 6]. On the other hand, Client—Server Databases provide high transaction throughput rates by effectively harnessing the resources available in a network of clients/workstations [22, 5, 4]. While algorithms for scheduling real-time tasks [14, 1, 8], real-time multiprocessor systems [19, 3, 15], and assignment of tasks in distributed environments [10] have received attention, there has been no coverage of real-time issues in the popular Client-Server database paradigm [11]. The goal of this chapter is to address the issue of real-time transaction processing in a Client-Server (CS) environment and provide preliminary performance indicators.