Abstract
A common pattern for enterprise applications, particularly in small and medium businesses, is the reliance on an integrated traditional relational database system that provides persistence and where the relational aspect underlies the core logic of the application. While several solutions are proposed for scaling out such applications, database replication is key if the relational aspect is to be preserved. However, it is worrisome that because proposed solutions for database replication have been evaluated using simple synthetic benchmarks, their applicability to enterprise applications is not straightforward: the performance of conservative solutions hinges on the ability to conveniently partition applications while optimistic solutions may experience unacceptable abort rates, compromising fairness, particularly considering long-running transactions. In this thesis, we address these challenges. First, by performing a detailed evaluation of the applicability of database replication protocols based on conservative concurrency control to enterprise applications. Results invalidate the common assumption that real-world databases can be easily partitioned. Then, we tackle the issue of unacceptable abort rates in optimistic solutions by proposing a novel transaction scheduler, AJITTS, which uses an adaptive mechanism that by reaching and maintaining the optimal level of concurrency in the system, minimizes aborts and improves throughput.;Um padrão comum no que toca a aplicações empresariais, particularmente em pequenas e médias empresas, é a dependência de um sistema de base dados relacional integrado que garante a persistência dos dados e no qual o aspeto relacional é parte integral da logica da aplicação. Embora várias soluções tenham sido propostas para dotar este tipo de aplicações de escalabilidade horizontal, a replicação de base de dados é a solução se o aspeto relacional deve ser preservado. No entanto, é preocupante que, dado que as soluções existentes para replicação de base de dados têm sido avaliadas utilizando testes de desempenho sintéticos e simples, a aplicabilidade destes a aplicações empresariais não é directa: o desempenho de soluções conservadoras está intimamente ligado à capacidade de particionar a aplicação convenientemente, enquanto que soluções optimistas podem sofrer de taxas de insucesso inaceitáveis o que compromete a equidade das mesmas, em particular no caso de transações especialmente longas. Nesta tese, abordamos estes desafios. Primeiro, através de uma avaliação detalhada da aplicabilidade de protocolos de replicação de base de dados baseados em controlo de concorrência conservador a aplicações empresariais. Os resultados obtidos invalidam o pressuposto comum de que bases de dados reais podem ser facilmente particionadas. Assim sendo, abordámos o problema das possíveis taxas de insucesso inaceitáveis em soluções optimistas propondo um novo escalonador de transações, o AJITTS, que utiliza um mecanismo adaptativo que ao atingir e manter o nível ótimo de concorrência no sistema, minimiza a taxa de insucesso e melhora o desempenho do mesmo.

Abstract
Typed models of connector/component composition specify interfaces describing ports of components and connectors. Typing ensures that these ports are plugged together appropriately, so that data can flow out of each output port and into an input port. These interfaces typically consider the direction of data flow and the type of values flowing. Components, connectors, and systems are often parameterised in such a way that the parameters affect the interfaces. Typing such connector families is challenging. This paper takes a first step towards addressing this problem by presenting a calculus of connector families with integer and boolean parameters. The calculus is based on monoidal categories, with a dependent type system that describes the parameterised interfaces of these connectors. We use families of Reo connectors as running examples, and show how this calculus can be applied to Petri Nets and to BIP systems. The paper focuses on the structure of connectors—well-connectedness—and less on their behaviour, making it easily applicable to a wide range of coordination and component-based models. A type-checking algorithm based on constraints is used to analyse connector families, supported by a proof-of-concept implementation.

Abstract

Abstract
This paper briefly presents a new approach for helping developers identify energy hot spots in their applications. Using tests cases, and statistical methods based on Spectrum-based Fault Localization, high energy consumption is related to the system's source code and a ranking of possible energy leaks are pointed. This technique is both language independent, and context level independent. Initial studies have also shown that using this technique helped developers identify and optimize energy problems in half the time while improving the energy efficiency by 18%. © 2017 IEEE.

Abstract
In recent years we witnessed the arrival of new trends, such as server virtualization and cloud services, an increasing number of mobile devices and online contents, leading the networking industry to deliberate about how traditional network architectures can be adapted or even deciding if a new perspective for them should be taken. SDN (Software-Defined Networking) emerged under this framing, opening a road for new developments due to the centralized logic control and view of the network, the decoupling of data and control planes, and the abstraction of the underlying network infrastructure from the applications. Although firstly oriented to packet switching, network measurements have also emerged as one promising field for SDN, as its flexibility enables programmable measurements, allowing a SDN controller to manage measurement tasks concurrently at multiple spatial and temporal scales. In this context, this paper is focused on exploring the SDN architecture and components for supporting the flexible selection and configuration of network monitoring tasks that rely on the use of traffic sampling. The aim is to take advantage of the integrated view of SDN controllers to apply and configure appropriate sampling techniques in network measurement points according to the requirements of specific measurement tasks. Through SDN, flexible and service-oriented configuration of network monitoring can be achieved, allowing also to improve the trade-off between accuracy and overhead of the monitoring process. In this way, this study, examining relevant SDN elements and solutions for deploying this monitoring paradigm, provides useful insights to enhance the programmability and efficiency of sampling-based network monitoring. © 2017, Springer International Publishing AG.

Abstract
Adaptability and energy-efficient sensing are essential properties to sustain the easy deployment and lifetime of WSNs. These properties assume a stronger role in autonomous sensing environments where the application objectives or the parameters under measurement vary, and human intervention is not viable. In this context, this paper proposes LiteSense, a self-adaptive sampling scheme for WSNs, which aims at capturing accurately the behavior of the physical parameters of interest in each WSN context yet reducing the overhead in terms of sensing events and, consequently, the energy consumption. For this purpose, a set of low-complexity rules auto-regulates the sensing frequency depending on the observed parameter variation. Resorting to real environmental datasets, we provide statistical results showing the ability of LiteSense in reducing sensing activity and power consumption, while keeping the estimation accuracy of sensing events.