Adjustable flow control filters and reflective memories as support for distributed real-time systems

Abstract

Distributed real-time systems are difficult to construct such that the system behaves predictably. In addition, many of these systems must operate in complex and uncertain environments. We discuss two main ideas for moving distributed real-time systems towards predictability: adjustable flow control filters and reflective memories. Both of these are hardware based solutions that should be exploited for enhancing predictability and robustness, especially as hardware becomes cheaper. Adjustable flow control filters is a new idea being proposed here, while reflective memories are available off-the-shelf.<>