Performance Analysis of a Middleware Demultiplexing Pattern

A key enabler of the recently adopted, assembly-centric development approach for distributed real-time software systems is QoS-enabled middleware, which provides reusable building blocks in the form of design patterns that codify solutions to commonly recurring problems. These patterns can be customized by choosing an appropriate set of configuration parameters. The configuration options of a pattern exert a strong influence on system performance, which especially for real-time systems is of paramount importance. Despite this significant influence, currently there are no techniques available to analyze performance at design time, prior to the use of a pattern in a system. Many software systems are based on an event-driven paradigm, primarily because it fosters evolvability and composability. The event demultiplexing and dispatching capabilities that are uniform across such systems are encapsulated in the reactor pattern, which can be used to facilitate their development. Design-time performance analysis of these event-driven systems thus requires a model of the reactor pattern. In this paper, we present a performance model of the reactor pattern based on the stochastic reward net (SRN) modeling paradigm. We discuss how the model can be used to obtain performance metrics such as throughput, loss probability and upper and lower bounds on the response time. We illustrate the use of the model to guide the selection of configuration options and for sensitivity analysis using a case study of a handheld mobile device. We also validate the performance estimates obtained from the model using simulation