Rakis: Secure Fast I/O Primitives Across Trust Boundaries on Intel SGX

The use of Intel® Software Guard Extensions (SGX) offers robust security measures for shielding applications in untrusted environments. However, the performance overhead experienced by IO-intensive applications within SGX limits widespread adoption. Prior approaches have proposed the use of userspace kernel-bypass libraries such as Data Plane Development Kit (DPDK) inside SGX enclaves to enable direct access to IO devices. However, these solutions often come at the cost of increasing the Trusted Computing Base (TCB) size, expanding the attack surface, and complicating deployment. In this paper, we introduce Rakis, a comprehensive system that securely enables SGX enclave programs to leverage fast IO Linux kernel primitives without modifying user applications. Rakis prioritizes the security of its TCB components by employing rigorous software testing and verification methods, embodying a security-by-design approach. Importantly, Rakis achieves performance advantages without sacrificing TCB size or introducing deployment intricacies and demonstrates significant improvements in benchmark tests with a 4.6x increase in UDP network throughput compared to state-of-the-art SGX enclave LibOS (Gramine-SGX). To demonstrate the practical applicability of Rakis, we evaluate its performance on four real-world programs showcasing an average performance improvement of 2.8x compared to Gramine-SGX across all workloads.

Authors

Institutions

Publication Details

Published
2025-03-26
DOI
https://doi.org/10.1145/3689031.3696090
Citations
3
Primary Topic
Security and Verification in Computing
Type
article
Field-Weighted Citation Impact
7.44
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Rakis: Secure Fast I/O Primitives Across Trust Boundaries on Intel SGX

Fan Sang, Mansour Alharthi, Dmitrii Kuvaiskii, Mona Vij et al.
3 citations
Security and Verification in Computing
7.44
article

Rakis: Secure Fast I/O Primitives Across Trust Boundaries on Intel SGX

Fan Sang, Mansour Alharthi, Dmitrii Kuvaiskii, Mona Vij, Taesoo Kim
article en
3 citations

Abstract

The use of Intel® Software Guard Extensions (SGX) offers robust security measures for shielding applications in untrusted environments. However, the performance overhead experienced by IO-intensive applications within SGX limits widespread adoption. Prior approaches have proposed the use of userspace kernel-bypass libraries such as Data Plane Development Kit (DPDK) inside SGX enclaves to enable direct access to IO devices. However, these solutions often come at the cost of increasing the Trusted Computing Base (TCB) size, expanding the attack surface, and complicating deployment. In this paper, we introduce Rakis, a comprehensive system that securely enables SGX enclave programs to leverage fast IO Linux kernel primitives without modifying user applications. Rakis prioritizes the security of its TCB components by employing rigorous software testing and verification methods, embodying a security-by-design approach. Importantly, Rakis achieves performance advantages without sacrificing TCB size or introducing deployment intricacies and demonstrates significant improvements in benchmark tests with a 4.6x increase in UDP network throughput compared to state-of-the-art SGX enclave LibOS (Gramine-SGX). To demonstrate the practical applicability of Rakis, we evaluate its performance on four real-world programs showcasing an average performance improvement of 2.8x compared to Gramine-SGX across all workloads.

Georgia Institute of Technology (US), Intel (United States) (US)
Openalex Percentile: Top 4%
Security and Verification in Computing
7.44
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.