Sistemas computacionais são constituídos por componentes de hardware e software que podem eventualmente falhar. Por esse motivo, o mecanismo de transação sempre foi imprescindível para a construção de sistemas robustos. O suporte transacional para a tecnologia Web services foi definido em agosto de 2005, num conjunto de três especificações denominadas WS-Coordination, WS-AtomicTransaction e WS-BusinessActivity. Juntas, essas especificações definem um alicerce sobre o qual aplicações robustas baseadas em Web services podem ser construídas. Nesta dissertação realizamos um estudo sobre transações atômicas em ambientes Web services. Em particular, estendemos o gerenciador de transações presente no servidor de aplicações JBoss, de modo que ele passasse a comportar transações distribuídas envolvendo Web services. Além disso, avaliamos o desempenho desse gerenciador de transações quando ele emprega cada um dos seguintes mecanismos de chamada remota: Web services/SOAP, CORBA/IIOP e JBoss Remoting. Finalmente, realizamos experimentos de escalabilidade e interoperabilidade. / Computing systems consist of a multitude of hardware and software components that may fail. For this reason, the transaction mechanism has always been essential for the development of robust systems. Transactional support for the Web services technology was defined in August 2005, in a set of three specifications, namely WS-Coordination, WS-AtomicTransaction, and WS-BusinessActivity. Together, such specifications enable the development of robust Web services applications. In this dissertation we studied atomic transactions in the Web services realm. Particularly, we added Web services atomic transaction support to the existing JBoss application server transaction manager. Furthermore, we evaluated the performance of this transaction manager when it employs each of the following remote method invocation mechanisms: Web services/SOAP, CORBA/IIOP and JBoss Remoting. Finally, we performed scalability and interoperability experiments.

Finding concurrency bugs in complex software is difficult. As a contribution to coping with this problem the thesis proposes an architecture for a fully automated dynamic detection and healing of data races and atomicity violations in Java. Two distinct algorithms for detecting of data races are presented. One of them is a novel algorithm called AtomRace which detects data races as a special case of atomicity violations. The healing is based on suppressing a recurrence of the detected problem and can be performed by introducing an additional synchronization or by legally influencing the Java scheduler. Basically forces certain parts of the code  to be executed atomically. The proposed architecture uses bytecode instrumentation to be able to track and influence the execution. The architecture and algorithms were implemented and tested on multiple case studies.