About the project
The fast technological evolution and the increased popularity of communication networks lead to the realization of services and distributed applications which are becoming more and more important in the Information Society. These services and applications have to be fast, reliable and secure. Hence, performance evaluation plays a critical role. The existing techniques and tools for performance evaluation have reached their limits in addressing the analysis and the design of the emerging information systems.
This Project focuses on the foundations of performance evaluation with the aim of developing new techniques, methodologies and tools for the analysis of the complex systems driving the Information Society. The topics addressed within the Project will encompass all the aspects of performance evaluation, from measurement and characterization techniques, to formalisms for the qualitative and quantitative modelling of complex systems, to solution methods for analytical, numerical and simulation models.
The research activities carried out within the Project can be subdivided into the following topics, corresponding to the Work Packages of the Project.
WP1: Measurement techniques and workload characterization
Within this topic, the Project plans to extend the measurement and workload characterization techniques of multi-layer complex systems. We will define methodologies for benchmarking and we will identify the types of information to be measured to describe the performance and the behavior of complex systems. Workload characterization studies will deal with techniques and models able to capture and reproduce the workload and the user behavior. Measurement techniques and workload characterization will be applied to identify the variables and the components involved in the energetic consumption of mobile terminals. Another application will deal with characterization of replication policies in terms of overhead induced on the system.
WP2: Formalisms and modelling
Within this topic, the Project plans to study models based on formalisms, such as, single queue, queueing networks, Markovian and non Markovian stochastic processes, timed Petri nets, non Markovian stochastic Petri nets, process algebra and fault tree. This research is aimed at extending the characteristics of the systems to be described within a model and to define new languages for describing the properties of the systems. We will define techniques for incremental and modular construction of the models. We will also address the modelling issues related to the systems with Quality of Service (QoS) constraints, with the objective of studying the risk of a provider subject to Service Level Agreement (SLA) contracts.
WP3: Automatic generation of predictive and validation models
Within this topic, the Project plans to study design techniques for a life-cycle validation of efficiency, reliability, safety and security. These techniques will allow designers to predict the ability of the system to meet the performance requirements and to validate user requirements before the implementation of the system itself. We will develop methodologies whose objective is to ease the construction of the models by means of an automatic translation process from formal specifications to performance models.
WP4: Solution techniques
Within this topic, the Project plans to study various types of techniques with the aim of modelling the performance of time-dependent systems by means of an integrated approach. This study will compare different types of models used to represent temporal constraints. These models will have to capture the statistical distribution and the deterministic duration (e.g., minimum and maximum constraints) of timed activities. We also plan to extend the solution techniques of stochastic and fluid Petri nets by using the composition principle of the Kronecker algebra, the symmetries of the models and the approximation via discrete models.
Within this topic, the Project plans to define and implement a platform for distributed simulation characterized by an efficient Runtime Infrastructure (RTI). We will implement, within a single software (service-layer), all the support services required to guarantee the correctness of the evolution of the simulation and good performance. We also plan to design and implement models based on distributed simulation aimed at the analysis and the evaluation of two different application domains, namely, the UMTS system characterized by heterogeneous applications (voice, video, data transfer, web browsing) and the Web systems with local and remote replications and guaranteed performance requirements even in presence of self-similar workload.
The experimentation on real systems of all the methodologies developed within the Project is an integral part of the activities. The information system of the Italian "Presidenza del Consiglio dei Ministri - Ufficio per l'Informatica, la Telematica e la Statistica" will be used as a test bed for the entire Project.