Benchmarking the Second Generation of Intel SGX Hardware

El-Hindi, Muhammad; Ziegler, Tobias; Heinrich, Matthias; Lutsch, Adrian; Zhao, Zheguang; Binnig, Carsten

doi:10.1145/3533737.3535098

Cited by 18 publications

(10 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…When a program runs inside a hardware enclave, the secure region of protected memory that it can use is called the Enclave Page Cache (EPC). As noted in our Challenges section, the EPC size in main memory is relatively small (128-256 MB [28]) compared to the standard main memory size (dozens of GBs). Datasets used in GWAS are typically hundreds of GB or even TBs in size, depending on the number of patients and SNPs.…”

Section: Resultsmentioning

confidence: 99%

“…In all our experiments for SECRET-GWAS, except for the EPC sensitivity test in Figure 3c, we did not trigger EPC paging. Recent work shows that increasing the scalability of the EPC is possible [35, 36], and Intel has developed new approaches to increase EPC memory in SGX [28] that have recently been added to Azure’s cloud offerings. Therefore, we believe that GWAS will be able to support even larger datasets with several million patients as the pace of EPC technology improves.…”

Section: Resultsmentioning

confidence: 99%

“…A larger EPC would grow the size of the Merkle tree and increase the already high cache-miss performance overhead (19.5% and 102% for writes and reads respectively) making it prohibitively expensive [40]. SGX, for instance, only provides 128-256 MB of EPC [28]. If an application’s working data set size exceeds EPC size, then it triggers EPC paging - an expensive process where encrypted pages are moved back and forth between non-enclave and enclave memory, slowing down programs by 5x on average [35].…”

Section: Methodsmentioning

confidence: 99%

See 2 more Smart Citations

SECRET-GWAS: Confidential Computing for Population-Scale GWAS

Rosenblum,

Dong,

Narayanasamy

2024

Preprint

View full text Add to dashboard Cite

Genomic data from a single institution lacks global diversity representation, especially for rare variants and diseases. Confidential computing can enable collaborative GWAS without compromising privacy or accuracy, however, due to limited secure memory space and performance overheads previous solutions fail to support widely used regression methods. We present SECRET-GWAS: a rapid, privacy-preserving, population-scale, collaborative GWAS tool. We discuss several system optimizations, including streaming, batching, data parallelization, and reducing trusted hardware overheads to efficiently scale linear and logistic regression to over a thousand processor cores on an Intel SGX-based cloud platform. In addition, we protect SECRET-GWAS against several hardware side-channel attacks, including Spectre, using data-oblivious code transformations and optimized speculative load hardening. SECRET-GWAS is an open-source tool and works with the widely used Hail genomic analysis framework. Our experiments on Azure's Confidential Computing platform demonstrate that SECRET-GWAS enables multivariate linear and logistic regression GWAS queries on population-scale datasets (one million patients, four million SNPs, 12 covariates) from ten independent sources in just 4.5 and 29 minutes, respectively.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

SECRET-GWAS: Confidential Computing for Population-Scale GWAS

Rosenblum,

Dong,

Narayanasamy

2024

Preprint

View full text Add to dashboard Cite

show abstract

“…Currently, the second generation of SGX includes a page cache with 128 MB. Data that does not fit on this cache must be swapped to/from an encrypted memory region, leading to a performance penalty [11, 8]. Also, in our setup, the amount of encrypted memory attributed to SGX is limited to 4GB in each server.…”

Section: Performance and Scalability Analysismentioning

confidence: 99%

“…Also, in our setup, the amount of encrypted memory attributed to SGX is limited to 4GB in each server. For memory-intensive GWAS algorithms, this means additional swapping to/from disk [24, 11]. Note that this SGX technological limitation cannot be solved by simply upgrading the server’s hardware but is addressed by G yosa ’s distributed design.…”

Section: Performance and Scalability Analysismentioning

confidence: 99%

A Distributed Computing Solution for Privacy-Preserving Genome-Wide Association Studies

Brito,

Ferreira,

Paulo

2024

Preprint

View full text Add to dashboard Cite

Breakthroughs in sequencing technologies led to an exponential growth of genomic data, providing unprecedented biological insights and new therapeutic applications. However, analyzing such large amounts of sensitive data raises key concerns regarding data privacy, specifically when the information is outsourced to third-party infrastructures for data storage and processing (e.g., cloud computing). Current solutions for data privacy protection resort to centralized designs or cryptographic primitives that impose considerable computational overheads, limiting their applicability to large-scale genomic analysis. We introduce GYOSA, a secure and privacy-preserving distributed genomic analysis solution. Unlike in previous work, GYOSA follows a distributed processing design that enables handling larger amounts of genomic data in a scalable and efficient fashion. Further, by leveraging trusted execution environments (TEEs), namely Intel SGX, GYOSA allows users to confidentially delegate their GWAS analysis to untrusted third-party infrastructures. To overcome the memory limitations of SGX, we implement a computation partitioning scheme within GYOSA. This scheme reduces the number of operations done inside the TEEs while safeguarding the users' genomic data privacy. By integrating this security scheme in Glow, GYOSA provides a secure and distributed environment that facilitates diverse GWAS studies. The experimental evaluation validates the applicability and scalability of GYOSA, reinforcing its ability to provide enhanced security guarantees. Further, the results show that, by distributing GWASes computations, one can achieve a practical and usable privacy-preserving solution.

show abstract

Privacy-Preserving Machine Learning on Apache Spark

Brito,

Ferreira,

Portela

et al. 2023

IEEE Access

View full text Add to dashboard Cite

The adoption of third-party machine learning (ML) cloud services is highly dependent on the security guarantees and the performance penalty they incur on workloads for model training and inference. This paper explores security/performance trade-offs for the distributed Apache Spark framework and its ML library. Concretely, we build upon a key insight: in specific deployment settings, one can reveal carefully chosen non-sensitive operations (e.g. statistical calculations). This allows us to considerably improve the performance of privacy-preserving solutions without exposing the protocol to pervasive ML attacks. In more detail, we propose Soteria, a system for distributed privacy-preserving ML that leverages Trusted Execution Environments (e.g. Intel SGX) to run computations over sensitive information in isolated containers (enclaves). Unlike previous work, where all ML-related computation is performed at trusted enclaves, we introduce a hybrid scheme, combining computation done inside and outside these enclaves. The experimental evaluation validates that our approach reduces the runtime of ML algorithms by up to 41% when compared to previous related work. Our protocol is accompanied by a security proof and a discussion regarding resilience against a wide spectrum of ML attacks.

show abstract

Benchmarking the Second Generation of Intel SGX Hardware

Cited by 18 publications

References 24 publications

SECRET-GWAS: Confidential Computing for Population-Scale GWAS

SECRET-GWAS: Confidential Computing for Population-Scale GWAS

A Distributed Computing Solution for Privacy-Preserving Genome-Wide Association Studies

Privacy-Preserving Machine Learning on Apache Spark

Contact Info

Product

Resources

About