Modelos de gerenciamento de enclaves para execução segura de componentes de software

• Ano de defesa: 2020
• Autor(a) principal: Will, Newton Carlos
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Paraná; Dois Vizinhos; Brasil; Pós-Graduação em Informática; UFPR
• Palavras-chave em Português: Arquitetura de software; Programação (Computadores); Computação de alto desempenho; Software architecture; Computer programming; High performance computing; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO; INFORMÁTICA (40001016034P5)
• Link de acesso: http://repositorio.utfpr.edu.br/jspui/handle/1/32122

Data confidentiality is becoming increasingly important to computer users, whether in a corporate environment or even in a home environment. Not only are business-sensitive data currently being trafficked across the network or being handled by a variety of software, but there is also an intense use of applications for banking transactions and other commonly used applications that manipulate sensitive user data, which must have their confidentiality and integrity guaranteed. In this sense, there are several solutions being proposed to maintain the confidentiality and integrity of the data, among them the Intel SGX (Software Guard Extensions) architecture, which has mechanisms to encapsulate applications and data in a protected area of memory having restricted access, making it impossible to access this region of memory to other applications or to the operating system. The use of such mechanisms to provide the confidentiality and integrity of sensitive data results in a performance impact during the application execution, due to the restrictions and verifications imposed by the Intel SGX architecture. The present work aims to analyze the programming models that are applied in solutions that use the Intel SGX architecture and present alternatives that seek a more efficient use of the resources provided by this architecture and also the reduction of the performance impact due to its use. Thus, two management models are presented: (i) enclave sharing; and (ii) enclave pool. In order to apply such models, an architecture of an enclave provider is proposed, which offers a decoupling between the enclave and the application that uses it, allowing to apply the proposed management models and offering the resources provided by the enclaves to the applications in “as a service” format. A prototype is built to evaluate the proposed architecture and models, with the performance tests demonstrating considerable reductions in the impact for enclave requests, while guaranteeing good response to attend simultaneous requests. Thus, it is concluded that the use of architectural software models can bring benefits in resource management and performance gains in the execution of secure applications.