DTX: um mecanismo de controle de concorrência distribuído para dados XML

• Ano de defesa: 2008
• Autor(a) principal: Moreira, Leonardo Oliveira
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Ciência da computação; XML; Transações distribuídas; Controle de concorrência; Distributed transactions; Concurrency control
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/18458

XML has become a widely used standard for data representation and exchange among Web applications. Consequently, a large volume of such data is distributed all over the Web and stored using several persistence methods. DBMSs provide concurrency control techniques to manage data. However, the structure of XML data makes it difficult to use these techniques. Projects are being proposed and they provide management of XML documents. Nevertheless, most of these projects do not provide efficient concurrency control mechanisms for distributed data. Some of them do provide support for distributed control of XML data, but use protocols that have limitations and offer low concurrency levels. In order to provide effective data management in distributed environments, we present DTX, a distributed concurrency control mechanism for XML data that takes into account its structural characteristics. DTX aims to provide effective management of XML data and contemplate properties such as isolation and consistency in transactions, using a multi-granular concurrency control protocol that increases parallelism among transactions and that has an optimized structure for data representation. DTX has a modular and flexible architecture, allowing for easy integration with any XML storage mechanisms. Moreover, DTX can be extended by adding new features to it. In order to evaluate DTX, several experiments were conducted comparing DTX as it is with a variation of DTX that uses a fine-grained protocol, in an attempt to simulate existing strategies in related work. Results confirm DTX’s effectiveness considering different aspects of distributed transactions on XML data, improving their performance, i.e., transaction execution time.