Xavier Franch

Systematic formulation of non-functional characteristics of software

This paper presents NoFun, a notation aimed at dealing with non-functional aspects of software systems at the product level in the component programming framework. NoFun can be used to define hierarchies of non-functional attributes, which can be bound to individual software components, libraries of components or (sets of) software systems. Non-functional attributes can be defined in several ways, being possible to choose a particular definition in a concrete context. Also, NoFun allows to state the values of the attributes in component implementations, and to formulate non-functional requirements over component implementations. The notation is complemented with an algorithm able to select the best implementation of components (with respect to their non-functional characteristics) in their context of use.

PRiM: An i * -based process reengineering method for information systems specification

Information system development can often be addressed as a business process reengineering practice, either because it automates some human-based processes or because it replaces an existing legacy system. Therefore, observing and analysing current processes can enable analysts to converge on the specification of the new system, generating and evaluating new system alternatives throughout. In this paper, we propose a method to support this reengineering process that analyses the strengths and weaknesses of the current process; considers the strategic needs of the organization; provides guidelines for the prescriptive construction of i^* models; and drives the systematic generation and evaluation of alternative technological and organizational solutions for the new system.

DesCOTS: a software system for selecting COTS components

Selection of commercial-off-the-shelf software components (COTS components) has a growing importance in software engineering. Unfortunately, selection projects have a high risk of ending up into abandonment or yielding an incorrect selection. The use of some software engineering practices such as the definition of quality models can reduce this risk. We defined a process for COTS components selection based on the use of quality models and we started to apply it in academic and industrial cases. The need of having a tool to support this process arose and, although some tools already exist to partially support the involved activities, none of them was suitable enough. Because of this we developed DesCOTS, a software system that embraces several tools that interact to support the different activities of our process. The system has been designed taking into account not only functional concerns but also nonfunctional aspects such as reusability, interoperability and portability. We present in this paper the different subsystems of DesCOTS and discuss about their applicability.

Sustainability in software engineering: A systematic literature review

Background: Supporting sustainability in software engineering is becoming an active area of research. We want to contribute the first Systematic Literature Review(SLR) in this field to aid researchers who are motivated to contribute to that topic by providing a body of knowledge as starting point, because we know from own experience, this search can be tedious and time consuming. Aim: We aim to provide an overview of different aspects of sustainability in software engineering research with regard to research activity, investigated topics, identified limitations, proposed approaches, used methods, available studies, and considered domains. Method: The applied method is a SLR in five reliable and commonly-used databases according to the (quasi-standard) protocol by Kitchenham et al. [1]. We assessed the 100 first results of each database ordered by relevance with respect to the search query. Results: Of 500 classified publications, we regard 96 as relevant for our research questions. We sketch a taxonomy of their topics and domains, and provide lists of used methods and proposed approaches. Most of the excluded publications were ruled out because of an unfitting usage of terms within the search query. Conclusions: Currently, there is little research coverage on the different aspects of sustainability in software engineering while other disciplines are already more active. Future work includes extending the study by reviewing a higher number of publications, including dedicated journal and workshop searches, and snowballing.

Dealing with Non-Functional Requirements in Model-Driven Development

The impact of non-functional requirements (NFRs) over software systems has been widely documented. Consequently, cost-effective software production method shall provide means to integrate this type of requirements into the development process. In this vision paper we analyze this assumption over a particular type of software production paradigm: model-driven development (MDD). We report first the current state of MDD approaches with respect to NFRs and remark that, in general, NFRs are not addressed in MDD methods and processes, and we discuss the effects of this situation. Next, we outline a general framework that integrates NFRs into the core of the MDD process and provide a detailed comparison among all the MDD approaches considered. Last, we identify some research issues related to this framework.

Service Level Agreement Monitor (SALMon)

One of the most successful architectural styles nowadays is service oriented architecture (SOA). In this type of architecture there are a lot of dependencies between services, but each service is an independent element of the system. In this situation we need some way to ensure that every service is working correctly and to take actions when something goes wrong to evolve the architecture as fast as we can. For example, if one of the lower level services of the service composition stops working, it could lead to a total or partial system malfunction. In this situation there is a need to be able to build reliable SOA systems. Our proposal, SALMon, is based on monitoring the services for service level agreement (SLA) violations. The SALMon architecture is composed of three types of components: Monitors that are composed of measure instruments, the measured quality attributes being taken from an ISO/IEC 9126-1-based service oriented quality model; analyzers that check the SLA rules; and decision makers that perform corrective actions to satisfy SLA rules again. These 3 types of components are mostly technology-independent and they act as services inside of a SOA system making our architecture very scalable and comfortable for its purpose.

Formalising ERP Selection Criteria

We present a proposal for selecting enterprise resource planning (ERP) products from a formal description of their relevant characteristics. The work is based on a previous and successfull collaboration with a midsize company in the field of software package selection. An ERP was selected following a systematic methodology called SHERPA. In past experience, SHERPA relied on natural language descriptions of the application domain, user needs and candidate ERP solutions. We show that a formal language may be used for modeling this application domain, translating user needs into requirements for the ERP products, and for reflecting how concrete ERP products adjust to them. Having selection criteria used during ERP acquisition formally modeled as well as user needs and ERP product descriptions, we expect to obtain more reliable and understandable results in this process.

On the quantitative analysis of agent-oriented models

Agent-oriented models are used in organization and information system modelling for providing intentional descriptions of processes as a network of relationships among actors. As such, they capture and represent goals, dependencies, intentions, beliefs, alternatives, etc., which appear in several contexts: business process reengineering, information system development, etc. In this paper, we are interested in the definition of a framework for the analysis of the properties that these models exhibit. Indicators and metrics for these properties are defined in terms of the model elements (e.g., actors, dependencies, scenario paths, etc.) Our approach is basically quantitative in nature, which allows defining indicators and metrics that can be reused in many contexts. However, a qualitative component can be introduced if trustable expert knowledge is available; the extent up to which quantitative and qualitative aspects are intertwined can be determined in every single case. We apply our proposal to the i* notation and we take as main case study a highly-intentional property, predictability of model elements.

Putting non-functional requirements into software architecture

This paper presents an approach for incorporating non-functional information of software system into software architectures. To do so, components present two distinguished slots: their non-functional specification, where non-functional requirements on components are placed, and their non-functional behaviour with respect to these requirements. Also, connector protocols may describe which non-functional aspects are relevant to component connections. We propose a notation to describe non-functionality in a systematic manner, and we use it to analyse two particular aspects of the meeting scheduler case study, user interaction and performance.

A quality-model-based approach for describing and evaluating software packages

Selection of software packages from user requirements is a central task in software engineering. Selection of inappropriate packages may compromise business processes and may interfere negatively in the functioning of the involved organization. Success of package selection is endangered because of many factors, one of the most important being the absence of structured descriptions of both package features and user quality requirements. In this paper, we propose a methodology for describing quality factors of software packages using the ISO/IEC quality standard as a framework. Following this standard, relevant attributes for a specific software domain are identified and structured as a hierarchy, and metrics for them are chosen. Software packages in this domain can then be described in a uniform and comprehensive way. Therefore, selection of packages can be ameliorated by transforming user quality requirements into requirements expressed in terms of quality model attributes. We illustrate the approach by presenting, in some depth, a quality model for the mail server domain.