DOP-MS: Serviço de offloading de dados usando uma arquitetura de microsserviços com suporte a anonimização de dados

• Ano de defesa: 2021
• Autor(a) principal: Silvestre, Vitória Regina Nicolau
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Mobile Cloud Computing; Offloading; Microsserviços
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/68716

Due to mobile devices’ growing presence in our daily routine, mobile applications are becoming increasingly complex, requiring more powerful processing capability and more extensive data storage, which characterizes a challenge when computational constraints of these devices are taken into account. The data offloading technique enables data migration into a remote environment, allowing (i) storage savings on the mobile device and (ii) sharing data among users. Several software infrastructures have been proposed to help the development of mobile applications with data offloading features. However, they lack essential features for data offloading, such as configurable data synchronization policy models, privacy mechanisms for offloaded data, and scalability and performance analyses. This work presents a solution to assist the development of mobile applications that use data migration, including contextual data, from mobile devices to a remote environment, based on a microservice architecture. In some scenarios (e.g., medical patient monitoring applications), data from different users may be used to infer new situations and understand their execution environment. The proposed solution here is called DOP-MS, a data offloading service using a microservice architecture with support for data anonymization. DOP-MS development is based on the evolution and integration of two previous works: COP and CAOS-MS. We conducted two groups of experiments: a proof of concept to validate the developed solution and performance and scalability tests to verify if a microservice architecture brought benefits related to performance and scalability for the proposed solution. As a result of these tests, we concluded that data offloading provides benefits in savings in storage mobile devices and creates new possibilities for inferring situations based on multiple users’ sharing data. The performance and scalability experiments showed that the microservice architecture provides better support for scalability and better performance as long the number of DOP-MS instances is provided. Finally, the work presents a statistical analysis from the data obtained during the tests performed.