An Overview of the STanford Real-time Information Processor (STRIP)

Abstract

We believe that the greatest growth potential for soft realtime databases is not as isolated monolithic databases but as components in open systems consisting of many heterogeneous databases. In such environments, the flexibility to deal with unpredictable situations and the ability to cooperate with other databases (often non-real-time databases) is just as important as the guarantee of stringent timing constraints. In this paper, we describe a database designed explicitly for heterogeneous environments, the STanford Realtime Information Processor STRIP, which runs on standard Posix Unix, is a soft real-time main memory database with special facilities for importing and exporting data as well as handling derived data. We will describe the architecture of STRIP, its unique features, and its potential uses in overall system architectures. 1 Project Goals The STanford Real-time Information Processor (STRIP) was designed with the following goals: 1. support for transactions and data with soft timing constraints, 2. high performance, 3. high availability, 4. the ability to share data with other components in open systems. The first goal, support of soft timing constraints, is implemented through two mechanisms: value function scheduling and maximum age constraints. A number of different scheduling algorithms are available, including earliest deadline first, highest value first, highest value density first, and custom scheduling algorithms. To support temporal constraints on the data, users can define maximum age requirements for views and for data accesses, where the age of data is defined to be the amount of time since it was last changed. The second goal, high performance, is achieved by using a main memory database. To provide high availability, a STRIP system can be reconfigured on the fly: Changes to This work was partially supported by the Telecommunications Center at Stanford University, by Hewlett Packard, and by Philips. y Stanford University Department of Comp