ions of software connectors” Proc. 3rd International Conf. on Performance Engineering, pp. 275-278, 2013. [14] L. Kapova, Configurable Software Performance Completions through Higher-Order Model PhD Thesis, Karlsruhe Institute of Technology, 2011. [15] M. Alhaj, D.C. Petriu, Using Aspects for Platform-Independent to Platform-Dependent Model Transformations, International Journal of Electrical and Computer Systems, Vol. 1, Issue 1, pp.35-48, 2012. [16] J. Kienzle, W. A. Abed, F. Fleurey, J. Jezequel, J. Klein, “Aspect-Oriented Design with Reusable Aspect Models”, Transactions on Aspect-Oriented Software Development, vol. 7, pp. 279 – 327, 2010.

A software design often does not describe the software infrastructure it will need to run, but a performance analysis must account for its effects. “Performance completions” represent the infrastructure and must be incorporated in the application performance model. This paper considers completions for middleware. It proposes a unified framework for describing all kinds of middleware in the Layered Queuing Network (LQN) model, based on a generic template and elaborations for middleware features. The template is applied to several common request-reply middleware systems. A process is given for building a new middleware completion model and for incorporating it into a LQN model.

In cloud computing, an "edge cloud" may be introduced close to some of the end users, to give faster service for very demanding applications. The transactions that require heavy processing capacity and longer processing times are seen as more suitable to be carried out at the "core" cloud. Parts in the core and edge may then have to communicate, introducing associated network latencies. An application should be deployed across edge and core with the aim to reduce the overall effect of network latencies, in order to meet end user response time goals. In this paper, we use a Layered Queueing Network performance model to explore the impact of network latency and some possible deployment choices on the responsiveness of an application called HCAT (Home Care Aides Technology). The evaluations show that the use of the edge cloud may cause performance degradation, rather than gain, for some kinds of applications.

With the advent of the new very powerful sensors, the WSNs have extended their scope from simple isolated networks to multiple networks interconnected through Internet.To model these new complex networks, a level of abstraction higher with respect to the one adopted by the simulation models of the original sensor networks is required. In this paper we describe an alternative approach to the one used in the original simulation models of WSN which is based on Queueing Network (QN) technique. This technique is particularly oriented to the solution of capacity planning problems.We demonstrate through simulations that the results provided by the proposed QN model are reliable by comparing them to the ones obtained by a similar model implemented with the widely adopted network simulator ns-2. The results show a very good agreement between the values provided by the two modeling techniques.