Desenvolvimento de aplicações com Intel SGX para computação confidencial: uma análise quantitativa e qualitativa

Abstract

Cloud computing presents security problems during data processing, since there are no efficient protection measures implemented in the execution of programs or even in the hardware in which it will be executed. One solution to this problem is the application of confidential computing, which uses methods to make data processing more secure. However, as it is recent, the techniques may present loss of performance. This paper attempts to assess the impact on development and performance of Intel SGX confidential computing technology. For this, two applications were created, in two versions each, a base version for control and a version using Intel SGX. The first application performs mathematical operations, receiving values outside the enclave and processing them inside. The second application consists of running a sorting algorithm within the enclave, in order to evaluate the processing of a large volume of data in the enclave. Measurements were carried out to evaluate the performance obtained in each of the scenarios. In the results of the first application, the base application had better performance, in the second, the application with Intel SGX had better performance. The execution of the experiments revealed several difficulties to adapt a program to support the technology, showing that the option of using it in consolidated applications must be extensively evaluated. However, its use can be considered for new applications.