標題: 固態硬碟結構偵測方法
Geometry Detection for Solid-State Drives
作者: 郭郡杰
Kuo, Chun-Chieh
張立平
Chang, Li-Pin
資訊科學與工程研究所
關鍵字: 固態硬碟;快閃記憶體;NAND Flash Memory;Solid State Drive;SSD
公開日期: 2008
摘要: 最近,以NAND flash memory為基礎的Solid State Drive (SSD)正快速的發展,並且已經實際運用於筆記型電腦、數位攝影機、UMPC等嵌入式產品中,更有多家業者提出以SSD為主要儲存裝置的伺服器解決方案。然而過去的二十年中,許多研究都致力於Hard Disk Drive (HDD),因此整個儲存系統架構都針對HDD的物理特性進行最佳化,如作業系統、檔案系統、應用程式甚至資料庫的設計等。但SSD並不單純只是存取速度較快的HDD;由於NAND flash memory與磁碟截然不同的物理特性,傳統對於HDD的最佳化方式並不能直接套用在SSD中,甚至可能會帶來反效果。因此在這篇論文中我們想藉由了解SSD內部的硬體架構以及其韌體管理演算法,來對整個SSD的效能表現進行完整的分析。並且對於明顯影響效能的SSD geometry設計出一系列的測試方法,讓使用者能從產品外部藉由存取測式取得各種影響效能表現的參數,如基本存取單位大小或最佳循序寫入單位大小等。在得到這些明顯影響效能表現的參數後,便可直接運用於各種對SSD的最佳化中,如檔案系統的檔案位置分佈或記憶體空間分配大小等,讓SSD優異的存取效能達到更完整的發揮。
The Solid State Drive (SSD) based on NAND flash memory is fast developed recently, and being used in embedded products like notebook, digital camera, and UMPC etc., and many manufactures also choose SSD as a suitable storage for database server. However, in past 20 years, many researches were focusing on the geometry optimization of HDD, for example improving the design of OS, files system, application, and database. SSD is not just a faster HDD, so the geometry optimization for HDD can not be totally used is SSD. In the paper, we want to know the inside hardware design and firmware management for SSD. Therefore we design a series of detections for SSD geometry, and let the users get the important elements affecting SSD performance. By using these elements, we can optimize SSD design, and take fully usage of SSD outstanding data accessing ability.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT079655560
http://hdl.handle.net/11536/43364
Appears in Collections:Thesis


Files in This Item:

  1. 556001.pdf