Reproduce “Experimental Study on System-Level Performance Impact of Read Disturbance in Modern SSDs” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yonggon Park, Hyunuk Cho, Onur Mutlu, Sungjin Lee, Jisung Park
An experimental study characterizes the system-level performance impact of read disturbance in modern NVMe SSDs, revealing significant performance degradation, particularly under sequential read workloads and increased SSD age, and demonstrating new read-disturbance-based performance attacks capable of reducing victim I/O bandwidth by over 70%.
Key insights
- The system-level performance impact of read disturbance is highly dependent on SSD-internal reliability management strategies, which vary considerably across different SSD models, often leading to significant performance fluctuations.
- Many modern SSDs do not proactively utilize idle times for read-disturbance management, contrary to common assumptions, which contributes to persistent or sudden performance drops.
- Current SSD I/O scheduling policies, which often prioritize fairness without distinguishing internal management traffic, make shared SSDs vulnerable to read-disturbance-based performance attacks.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.