電機資訊學院: 資訊工程學研究所指導教授: 郭大維; 張原豪王澤元Wang, Tse-YuanTse-YuanWang2017-03-032018-07-052017-03-032018-07-052016http://ntur.lib.ntu.edu.tw//handle/246246/275550快閃記憶體的市場趨向顯示有越來越多的雲端儲存系統採用快閃記憶體當作儲存裝置，基於快閃記憶體的雲端儲存系統面臨必須滿足服務品質的挑戰，例如提供可預測的回應時間。然而，由於快閃記憶體的特性，當快閃記憶體儲存裝置觸發垃圾回收指令時，可能導致上層指令無法被立即的服務，這個現象導致不可預期的效能下降，產生回應時間差異過大的問題，因此如何最小化反應時間的差異性對基於快閃記憶體的雲端儲存系統是相當重要的議題，基於上述觀察，為了解決反應時間的差異性過大的問題，我們提出一個以配對為導向來減少快閃記憶體儲存裝置反應時間之最長的設計，最後，藉由實驗結果，我們驗證了所提出方法能有效減少快閃記憶體儲存裝置的最大反應時間。The market trend of flash memory shows that more and more flash storage devices are widely used in cloud storage systems. Flash-based storage servers face the challenge on fulfilling the requirement of quality of service (QoS); for example, offering predictable response time. However, due to the characteristics of flash memory, host commands are unable to process promptly when garbage collection is triggered. This phenomenon leads to unexpected performance degradation. Therefore, the critical issue for flash-based storage servers is to minimize the response-time variation. Based on the above observations, in order to resolve response-time variation problem, we propose a partnership design to minimize the maximal response time. Experimental results show that the proposed design is very effective in reducing the maximal response time of flash-memory storage devices.論文使用權限: 不同意授權儲存系統快閃記憶體效能垃圾回收配對Storage systemFlash memoryPerformanceGarbage collectionPartnershipthesis10.6342/NTU201600812