The case for aspect-oriented reflective middleware

Abstract

The emergence of applications domains such as pervasive and autonomic computing has increased the need for customisation and dynamic adaptation of both distributed systems, and the underlying middleware platforms. Two highly complementary technologies have been advocated to meet these challenges, namely: aspect oriented programming (AOP) and reflective middleware. However, these have so far been considered in isolation, or typically target a particular middleware challenge e.g. using aspects to customise a middleware implementation; or using reflection (or dynamic AOP) to alter runtime behaviour. We believe that in combination these technologies better support the engineering of dynamic distributed systems. In this paper, we explore how aspects and reflection have been utilised in both the programming language and middleware communities; building upon this work, we identify four core relationships that form the basis of our model for aspect-oriented reflective middleware. We then explore the potential of this model to i) increase support for the engineering of dynamic reconfigurations, and ii) improve the performance of adaptive systems.

Augmenting reflective middleware with an aspect orientation support layer

“…Turning now to related work, [10] contains general arguments for the benefits of combining reflection and AOP. With respect to this, our work can be seen as a case study or exemplar or evaluation of the more abstract perspective given in that paper.…”