Publicação

Run-time switching between total order algorithms

Ver documento

Detalhes bibliográficos
Resumo:A total order protocol is a fundamental building block in the construction of many distributed fault-tolerant applications. Unfortunately, the implementation of such a primitive can be expensive both in terms of communication steps and of number of messages exchanged. This problem is exacerbated in large-scale systems, where the performance of the algorithm may be limited by the presence of high-latency links. Optimistic total order protocols have been proposed to alleviate this problem. However, different optimistic protocols offer quite distinct services. Moreover, there are certain algorithms that perform better in specific scenarios and given network properties. This dissertation provides an overview of different optimistic approaches and establishes a characterization of their properties and suitability to different execution environments. An adaptive protocol that is able to dynamically switch between different total order algorithms is proposed and evaluated. The protocol allows to achieve the best possible performance by supporting the reconfiguration such that, in each moment, the algorithm that is most appropriate to the present network conditions can be executed. Experimental results show that, using our protocol, adaptation can be achieved with negligible interference in the data flow.
Autores principais:Mocito, José Carlos Vitório
Assunto:Total order Optimistic total order Adaptive algorithm Teses de mestrado - 2006
Ano:2006
País:Portugal
Tipo de documento:dissertação de mestrado
Tipo de acesso:acesso restrito
Instituição associada:Universidade de Lisboa
Idioma:inglês
Origem:Repositório da Universidade de Lisboa
Descrição
Resumo:A total order protocol is a fundamental building block in the construction of many distributed fault-tolerant applications. Unfortunately, the implementation of such a primitive can be expensive both in terms of communication steps and of number of messages exchanged. This problem is exacerbated in large-scale systems, where the performance of the algorithm may be limited by the presence of high-latency links. Optimistic total order protocols have been proposed to alleviate this problem. However, different optimistic protocols offer quite distinct services. Moreover, there are certain algorithms that perform better in specific scenarios and given network properties. This dissertation provides an overview of different optimistic approaches and establishes a characterization of their properties and suitability to different execution environments. An adaptive protocol that is able to dynamically switch between different total order algorithms is proposed and evaluated. The protocol allows to achieve the best possible performance by supporting the reconfiguration such that, in each moment, the algorithm that is most appropriate to the present network conditions can be executed. Experimental results show that, using our protocol, adaptation can be achieved with negligible interference in the data flow.