Security, Performance and Energy Trade-Offs of Hardware-Assisted Memory Protection Mechanisms

Abstract

The deployment of large-scale distributed systems, e.g., publish-subscribe platforms, that operate over sensitive data using the infrastructure of public cloud providers, is nowadays heavily hindered by the surging lack of trust toward the cloud operators. Although purely software-based solutions exist to protect the confidentiality of data and the processing itself, such as homomorphic encryption schemes, their performance is far from being practical under real-world workloads.The performance trade-offs of two novel hardware-assisted memory protection mechanisms, namely AMD SEV and Intel SGX -currently available on the market to tackle this problem, are described in this practical experience.Specifically, we implement and evaluate a publish/subscribe use-case and evaluate the impact of the memory protection mechanisms and the resulting performance. This paper reports on the experience gained while building this system, in particular when having to cope with the technical limitations imposed by SEV and SGX.Several trade-offs that provide valuable insights in terms of latency, throughput, processing time and energy requirements are exhibited by means of micro-and macro-benchmarks.

Faster enclave transitions for IO-intensive network applications

“…VM-based TEE implementations (e.g. AMD SEV, IBM SVM, and Intel TDX) support portability of legacy applications with a modest performance overhead [6], but have a larger attack surface and are vulnerable to several classes of attacks [17]. Process-based TEEs (e.g.…”

“…Intel SGX and ARM TrustZone) on the other hand have a smaller attack surface and improved security and were proposed for protecting network applications [12,13,24,30]. Unfortunately, the additional security checks together with memory access limitations negatively affect the performance of process-based TEEs [6]. Furthermore, these have shown to be particularly vulnerable to microarchitectural attacks [26] and platform vendors have repeatedly issued microcode patches to alleviate security problems [5].…”

Speeding up enclave transitions for IO-intensive applications

“…VM-based TEE implementations (e.g. AMD SEV, IBM SVM, and Intel TDX) support portability of legacy applications with a modest performance overhead [8], but have a larger attack surface and are vulnerable to several classes of attacks [18]. Process-based TEEs (e.g.…”

“…Intel SGX and ARM TrustZone) on the other hand have a smaller attack surface and improved security. Unfortunately, the additional security checks together with memory access limitations also affect the performance of process-based TEEs negatively [8]. Furthermore, these have shown to be particularly vulnerable to microarchitectural attacks [26] and platform vendors have repeatedly issued microcode patches to alleviate security problems [7].…”

Secure Stream Processing for Medical Data

“…Homomorphic encryption (HE) is a cryptographic scheme that allows evaluating algorithms over encrypted data without having to decrypt it [4]. In spite of reducing the trusted computing base (TCB) to zero on the remote side, HE frameworks are currently prohibitively slow [5]. On the other hand, we have Trusted Execution Environments (TEE).…”