Orthrus: Efficient and Timely Detection of Silent User Data Corruption in the Cloud with Resource-Adaptive Computation Validation

Chenxiao Liu, Zhenting Zhu(co-first), Quanxi Li, Yanwen Xia, Yifan Qiao, Xiangyun Deng, Youyou Lu, Tao Xie, Huimin Cui, Zidong Du, Harry Xu, Chenxi Wang*

SOSP 2025 (The 31st Symposium on Operating Systems Principles)

Detecting silent data corruption in cloud applications through computation validation.

My partBuilt the LLVM-based fault-injection infrastructure and an automated Python fault-injection platform.

Public code, environment setup, and evaluation instructions are available in the linked repositories.

Unmodified public PDF · UCLA author copy ↗ · CC BY 4.0

Introduction:

Even with substantial endeavors to test and validate processors, computational errors may still arise post-installation. One particular category of CPU errors transpires discreetly, without crashing applications or triggering hardware warnings. These elusive errors pose a significant threat by undermining user data, and their detection is challenging. This paper introduces Orthrus, a solution for the timely detection of silent user-data corruption caused by post-installation CPU errors.