No sugar but all the taste! Memory encryption without architectural support

12Citations
Citations of this article
15Readers
Mendeley users who have this article in their library.

This article is free to access.

Abstract

The protection of in situ data, typically require solutions that involve different kinds of encryption schemes. Even though the majority of these solutions prioritize the protection of cold data stored on secondary devices, it has been shown that sensitive information like passwords, secrets, and private data can be easily exfiltrated from main memory as well, by adversaries with physical access. As such, the protection of hot data that reside on main memory is equally important. In this paper, we aim to investigate whether it is possible to achieve memory encryption without any architectural support at a reasonable performance cost. In particular, we propose the first of its kind software-based memory encryption approach, which ensures that sensitive data will remain encrypted in main memory at all times. Our approach is based on commodity off-the-shelf hardware, and is totally transparent to legacy applications. To accommodate different applications needs, we have built two versions of main memory encryption: Full and Selective Memory Encryption. Additionally, we provide a new memory allocation library that allows programmers to manage granular sensitive memory regions according to the specific requirements of each application. We conduct an extensive quantitative evaluation and characterization of the overheads of our software-based memory encryption, using both micro-benchmarks and real-world application workloads. Our results show that the performance overheads due to memory encryption are tolerable in real-world network scenarios, below 17% for HTTP and 27% for HTTPS.

Cite

CITATION STYLE

APA

Papadopoulos, P., Vasiliadis, G., Christou, G., Markatos, E., & Ioannidis, S. (2017). No sugar but all the taste! Memory encryption without architectural support. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 10493 LNCS, pp. 362–380). Springer Verlag. https://doi.org/10.1007/978-3-319-66399-9_20

Register to see more suggestions

Mendeley helps you to discover research relevant for your work.

Already have an account?

Save time finding and organizing research with Mendeley

Sign up for free