CoordenaÃÃo de AperfeiÃoamento de Pessoal de NÃvel Superior / In clusters and virtualized data centers, the resources must be managed effectively by maximising the SLA fulfilment while minimising the cost. This work proposes a strategy for dynamic resource allocation in virtualized computing environments in order to reduce energy consumption without compromising performance requirements concerning availability and SLA violation. The proposed algorithms, based on classical heuristics, perform virtual machines migration between distinct hosts according to the variation in resources demand. These algorithms were evaluated from measurements in a real environment composed by heterogeneous virtualized hosts. We evaluate their performance in four different scenarios based on the CPU and memory utilization, number of migrations and energy consumption. In general, the results show that the algorithms responsible for consolidation and distribution of virtual machines between hosts are able to reduce energy consumption and dissipate the idle and overload points. / Em clusters e data centers virtualizados, os recursos precisam ser gerenciados com eficÃcia na busca de um trade-off entre a garantia de um atendimento satisfatÃrio à demanda por Qualidade de ServiÃo (QoS) e a reduÃÃo dos custos operacionais por parte dos provedores. Este trabalho propÃe uma estratÃgia para alocaÃÃo dinÃmica de recursos em ambientes computacionais virtualizados com vistas à reduÃÃo do consumo de energia, sem promover sobrecargas que podem comprometer o desempenho dos serviÃos ofertados. Os algoritmos propostos, baseados em heurÃsticas clÃssicas, realizam migraÃÃo de mÃquinas virtuais entre servidores distintos conforme variaÃÃo na demanda por recursos. Estes algoritmos foram verificados e validados por mediÃÃes em um ambiente real composto por servidores virtualizados heterogÃneos. O desempenho da proposta à avaliado em quatro cenÃrios distintos a partir das mÃtricas utilizaÃÃo de CPU, utilizaÃÃo de memÃria, nÃmero de migraÃÃes e consumo de energia. Os resultados mostraram que os algoritmos responsÃveis pela consolidaÃÃo e distribuiÃÃo das mÃquinas virtuais sÃo capazes de reduzir o consumo de energia e dissipar os pontos de Ãcio e sobrecarga do ambiente.

In computer security, cryptographic algorithms and protocols are required to ensure security of data and applications. This research investigates techniques to reduce the energy consumed by cryptographic hash functions. The specific hash functions considered are Message Digest-2 (MD2), Message Digest-5 (MD5), Secure Hash Algorithm-1 (SHA-1) and Secure Hash Algorithm-2 (SHA-2). The discussion around energy conservation in handheld devices like laptops and mobile devices is gaining momentum. Research has been done at the hardware and operating system levels to reduce the energy consumed by these devices. However, research on conserving energy at the application level is a new approach. This research is motivated by the energy consumed by anti-virus applications which use computationally intensive hash functions to ensure security. To reduce energy consumption by existing hash algorithms, the generic energy complexity model, designed by Roy et al. [Roy13], has been applied and tested. This model works by logically mapping the input across the eight available memory banks in the DDR3 architecture and accessing the data in parallel. In order to reduce the energy consumed, the data access pattern of the hash functions has been studied and the energy complexity model has been applied to hash functions to redesign the existing algorithms. These experiments have shown a reduction in the total energy consumed by hash functions with different degrees of parallelism of the input message, as the energy model predicted, thereby supporting the applicability of the energy model on the different hash functions chosen for the study. The study also compared the energy consumption by the hash functions to identify the hash function suitable for use based on required security level. Finally, statistical analysis was performed to verify the difference in energy consumption between MD5 and SHA2.

The Internet is rapidly growing in complexity, with increasing negative environmental and social impact. While heating and lighting are tangible examples of energy consumption, internet usage is not perceived as such. Therefore, it opens up opportunities for new, energy-efficient, slower, resource-saving and mindful protocols for the Internet to emerge. I propose Glow OS, an operating system that enables individuals and communities to align their online activities with intermittent solar energy. This system aims to accelerate the transition to a fossil-free internet by promoting solar literacy in the spirit of joy.