Inter-cloud is a recently emerging approach that expands cloud elasticity. By facilitating an adaptable setting, it purposes at the realization of a scalable resource provisioning that enables a diversity of cloud user requirements to be handled efficiently. This study’s contribution is in the inter-cloud performance optimization of job executions using metascheduling concepts. This includes the development of the inter-cloud meta-scheduling (ICMS) framework, the ICMS optimal schemes and the SimIC toolkit. The ICMS model is an architectural strategy for managing and scheduling user services in virtualized dynamically inter-linked clouds. This is achieved by the development of a model that includes a set of algorithms, namely the Service-Request, Service-Distribution, Service-Availability and Service-Allocation algorithms. These along with resource management optimal schemes offer the novel functionalities of the ICMS where the message exchanging implements the job distributions method, the VM deployment offers the VM management features and the local resource management system details the management of the local cloud schedulers. The generated system offers great flexibility by facilitating a lightweight resource management methodology while at the same time handling the heterogeneity of different clouds through advanced service level agreement coordination. Experimental results are productive as the proposed ICMS model achieves enhancement of the performance of service distribution for a variety of criteria such as service execution times, makespan, turnaround times, utilization levels and energy consumption rates for various inter-cloud entities, e.g. users, hosts and VMs. For example, ICMS optimizes the performance of a non-meta-brokering inter-cloud by 3%, while ICMS with full optimal schemes achieves 9% optimization for the same configurations. The whole experimental platform is implemented into the inter-cloud Simulation toolkit (SimIC) developed by the author, which is a discrete event simulation framework.

The increasing needs for computational power in areas such as weather simulation, genomics or Internet applications have led to sharing of geographically distributed and heterogeneous resources from commercial data centers and scientific institutions. Research in the areas of utility, grid and cloud computing, together with improvements in network and hardware virtualization has resulted in methods to locate and use resources to rapidly provision virtual environments in a flexible manner, while lowering costs for consumers and providers. However, there is still a lack of methodologies to enable efficient and seamless sharing of resources among institutions. In this work, we concentrate in the problem of executing parallel scientific applications across distributed resources belonging to separate organizations. Our approach can be divided in three main points. First, we define and implement an interoperable grid protocol to distribute job workloads among partners with different middleware and execution resources. Second, we research and implement different policies for virtual resource provisioning and job-to-resource allocation, taking advantage of their cooperation to improve execution cost and performance. Third, we explore the consequences of on-demand provisioning and allocation in the problem of site-selection for the execution of parallel workloads, and propose new strategies to reduce job slowdown and overall cost.

Orientador: Prof. Dr. Gustavo Sousa Pavani / Dissertação (mestrado) - Universidade Federal do ABC, Programa de Pós-Graduação em Ciência da Computação, 2017. / Algoritmos baseados em Otimização por Colonia de Formigas (Ant Colony Optimization { ACO) vem sendo usados com bastante sucesso no metaescalonamento distribuído e integrado dos recursos de computação e de comunicação em ambientes de grade computacional baseadas em redes opticas (lambda grid ). Nesse ambiente sao utilizados infraestruturas de comunicações compostas por enlaces de fibras opticas e nós opticos, que são elementos sujeitos aos mais variados tipos de falhas. Com efeito, problemas decorrentes dos equipamentos ou meios de transmissão podem interromper o trafego de informações e, consequentemente, causam a indisponibilidade de recursos na grade. A arquitetura proposta em [1] é capaz de gerenciar dinamicamente e de forma conjunta os recursos de rede e processamento no ambiente de lambda grid, além de prover agendamento e reserva futura desses recursos. Neste trabalho, considerar-se-à tambem a capacidade de sobrevivencia da grade sob condições adversas de falhas. Para tanto, a grade computacional deve prover mecanismos de restauração de forma a se recuperar em caso de falhas de enlace ou nó. Simulações foram realizadas com os diversos algoritmos de metaescalonamento propostos originalmente em [1], demonstrando o respectivo desempenho em termos de capacidade de restauração e de atraso de execução das tarefas restauradas. / Algorithms based on Ant Colony Optimization (ACO) have been successfully used in distributed and integrated meta-scheduling of computing and networking resources in lambda grids. The lambda grid environment is composed by optical fiber links and optical nodes, which are susceptible to diferent types of failure. In eect, problems due to equipment or transmission outages may interrupt the information trac and, consequently, cause unavailability of grid resources. The architecture proposed in [1] is capable of the dynamic, joint management of networking and processing resources at the lambda grid. It can also provide scheduling and advance reservation of those resources. In this work, we also consider the survivability capacity of the lambda grid when a failure occurs. Therefore, the lambda grid has to provide a restoration mechanism in order to recover from link and node failures. Simulations carried with the meta-scheduling algorithms originally proposed in [1] demonstrate their performance in terms of restorability and delay in scheduling the restored tasks.

No description available.