Abstract. Constructing and executing distributed applications that can adapt to their current operating context, in order to maintain or enhance Quality of Service (QoS) attribute levels, are complex tasks. Managing multiple, interacting QoS features is particularly difficult since these features tend to be distributed across the system and tangled with other features. The crosscutting nature of QoS features can make them difficult to evolve, and it can make it complicated to dynamically optimize with respect to provided QoS during execution. Furthermore, it complicates efficient construction of application variants that differ in their QoS characteristics to suit various execution contexts. This paper presents an aspect-oriented and model driven approach for constructing and a QoSaware middleware for execution of QoS-sensitive applications. Aspect-oriented modeling techniques are used to separate QoS features from primary application logic, and for efficient specification of alternative application variants. Model driven engineering techniques are used to derive run-time representations of application variants from platform independent models. The developed middleware chooses the best variant according to the current operating context and the available resources.
State-of-the-art dynamic middleware uses information about the environment in order to evaluate alternative configurations of an application and select one according to some criteria. In the context of applications sensitive to Quality of Service, we have identified the need for a platform independent description of configurations that includes non-functional behaviour, and that allows handling of a large number of application configurations. In this paper, we present a modelling principle and a service plan concept, which together represents such a description. The modelling principle and plan concept extend state-of-the-art with i) a model of the alternative configurations that ensure a minimum of reconfiguration steps; ii) a specification that contains information elements of the configuration, dependencies to the environment, and QoS characteristics; and iii) a platform independent specification. In the paper, we also perform a qualitative assessment of our approach, and we describe a proof-of-concept implementation.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.