DPAS: A Prompt, Accurate and Safe I/O Completion Method for SSDs
Dongjoo Seo, Jihyeon Jung, Yeohwan Yoon, Ping-Xiang Chen, Yongsoo Joo, Sung-Soo Lim, Nikil D. Dutt
摘要
Modern SSDs demand faster I/O completion methods. While polling is a potential alternative to interrupts, it suffers under CPU contention. Hybrid polling mitigates this by sleeping early and polling later, yet it cannot keep up with rapidly varying I/O latencies and incurs context-switch overheads. We introduce PAS, an accurate latency tracking method for hybrid polling that adjusts sleep duration using the two most recent I/Os, and DPAS, which dynamically switches among polling, interrupts, and PAS to overcome the inherent drawbacks of hybrid polling. Experiments show that PAS reduces CPU usage by 21 percentage points compared to Linux hybrid polling for 4 KB random reads, and DPAS improves YCSB performance by 9% on a 3D XPoint SSD and 5% on a TLC NAND SSD, even under simultaneous CPU contention and I/O interference.
问问这篇 Paper
智能体会读完全文。
Lune 把这篇 Paper 索引到了最后一个公式,引用它的顶会 Paper 也一样。你提问,回答直接引用原文。
它引用的顶会 Paper4
- LinnOS: Predictability on Unpredictable Flash Storage with a Light Neural NetworkMingzhe Hao, Levent Toksoz, Nanqinqin Li, Edward Edberg Halim 等OSDI 2020 · 被引用 97 次
- Baleen: ML Admission & Prefetching for Flash CachesDaniel Lin-Kit Wong, Hao Wu, Carson Molder, Sathya Gunasekar 等FAST 2024 · 被引用 26 次
- rkt-io: a direct I/O stack for shielded executionJörg Thalheim, Harshavardhan Unnibhavi, Christian Priebe, Pramod Bhatotia 等EuroSys 2021 · 被引用 24 次
- Optimizing Storage Performance with Calibrated InterruptsAmy Tai, Igor Smolyar, Michael Wei, Dan TsafrirOSDI 2021
相关 Paper
- UnICom: A Universally High-Performant I/O Completion Mechanism for Modern Computer SystemsRiwei Pan, Yu Liang, Sam H. Noh, Lei Li 等FAST 2026 · 被引用 1 次
- DEAR: Improving Performance and Lifetime of SSDs Using Dynamic Error-Aware RefreshJaeyong Lee, Beomjun Kim, Myoungjun Chun, Myungsuk Kim 等MICRO 2025 · 被引用 2 次
- Write-Aware Timestamp Tracking: Effective and Efficient Page Replacement for Modern HardwareDemian E. Vöhringer, Viktor LeisVLDB 2023 · 被引用 11 次
- Frequent background polling on a shared thread, using light-weight compiler interruptsNilanjana Basu, Claudio Montanari, Jakob ErikssonPLDI 2021 · 被引用 4 次
- PA-Tree: Polled-Mode Asynchronous B+ Tree for NVMeLi Wang, Zining Zhang, Bingsheng He, Zhenjie ZhangICDE 2020 · 被引用 9 次
