Processamento confiável no ambiente operacional Seljuk-Amoeba.

Abstract

Building distributed applications is not a simple task. Designers of such systems have followed two complementary approaches to reduce design complexity, namely: (i) the use of appropriate development tools; and ii) the choice of the most restrictive failure semantics possible for the components that make up the system execution layer. The Seljuk model uses these two approaches to specify a structured way to provide fault tolerance services in the context of distributed operating environments, thus facilitating the construction and execution of applications with reliable operation requirements. In this paper, following the Seljuk model, and taking the distributed operational system Amoeba as a substrate, we present the design and implementation of the reliable Seljuk-Amoeba processing service. Our proposal is based on extending the functionality of Amoeba's processing service, creating a reliable processing service. Our goal is achieved through the introduction of an execution service and the development of redundancy management protocols in Amoeba's communication service. The new runtime service performs the traditional functions of Amoeba's runtime service and also provides mechanisms for creating replicated processing units.