隨著實體記憶體的需求不斷得增加，綠能儲存系統變成近年來的熱門話題，
由於資料量不斷的擴增，節能儲存系統變成當今儲存技術的重要議題。在分散式
系統中，硬碟的idle 時間、較差的隨機存取效能都會導致硬碟消耗電能，因此，
我們提出了一種寫入快取機制稱做CacheRAID，CacheRAID 的目的是用來改善
在RAID 儲存系統的隨機存取問題並且延長硬碟的idle 時間，讓RAID 硬碟進入
省電模式。
系統的可靠度與容錯機制也是儲存系統中重要得議題，在容錯系統方面有許
多重要的問題，首先像是當有資料遺失時儲存系統的容錯機制需要花多少時間來
回復資料?或是儲存系統可以容忍多少顆硬碟損壞?舉例而言，RAID 儲存系統通
常可以容忍一顆硬碟故障，換句話說，當有兩顆硬碟損壞時，RAID 儲存系統就
無法回復資料。最後，擁有較高容錯度的系統會伴隨著空間使用率問題，例如高
容錯率往往需要更多的data parity 區塊，而這些區塊會佔據許多硬碟的空間，因
此如何使用較少的空間成本來回復儲存系統也是個重要的議題。

Green storage is a hot topic at the moment with the demand of physical storage devices increasing. Because of the data size increasing, conserving energy of storage systems is becoming a growing concern in current storage technology. The factors of disk power consumption include disk idle time, poor random writing performance and random read in distributed file systems. Hence, we present an adaptive write cache mechanism - CacheRAID. Redundant Arrays of Inexpensive Disk (RAID) is widely used in modern distributed storage systems. Our CacheRAID aims to improve the random access problems that implicitly exist in RAID techniques to create more idle time of hard drives, and conserve RAID disk array energy.
Good reliability and space efficiency are quality requisites to a good fault tolerance mechanism. And there are some critical issues in fault tolerance system. First, how fast can fault tolerance system to rebuild their data block when storage system is failure? Then, the storage system is still recoverable under the upper bound limited of disk failure. For example, RAID system is usually recoverable when one disk is failed. On the other hand, the RAID storage system can’t rebuild itself when the numbers of failure disk are more than one. Finally, the space efficiency that is the trade of high fault tolerance factor. High fault tolerance usually need more data parity block, and these parities block will occupy much storage space in disk. Hence, how can we use less data to recovery our storage system? That is an important issue.

1. 介紹
2. 系統背景
2.1 相關系統研究
2.2 Log 分析
2.3 系統可靠度模型
3. CacheRAID 系統架構
3.1 系統架構
3.2 資料擺放演算法
3.3 系統可靠度分析
3.4 容錯機制設計
3.5 MD5 驗證
4. 實驗結果
4.1 實驗環境
4.2 實驗數據結果
4.2.1 能源消耗量比較
4.2.2 系統可靠度比較
5. 結論
6. 參考資料

[1] D.A. Patterson, G. Gibson, and R.H. Katz. A case for redundant arrays of inexpensive disks (RAID), volume 17. ACM, 1988.
[2] Sun Zhizhuo Li Xiao, Tan Yu-An. "Semi-raid: A reliable energy aware raid data layout for sequential data access." In MSST'11, IEEE 27th Symposium on Mass Storage Systems and Technologies, 2011.
[3] Dennis Colarelli,Dirk Grunwald. "Massive arrays of idle disks for storage archives". Proceeding Supercomputing '02 Proceedings of the 2002 ACM/IEEE conference on Supercomputing.
[4] Satoshi Hikida, Hieu Hanh Le, Haruo Yokota."A Power Saving Storage Method That Considers Individual Disk Rotation".17th International Conference, DASFAA 2012.
[5] Dong Li,Jun Wang."EERAID: energy efficient redundant and inexpensive disk array".Proceeding EW 11 Proceedings of the 11th workshop on ACM SIGOPS European workshop.
[6] Charles Weddle,Mathew Oldham,Jin Qian,An-I Andy Wang,Peter Reiher,Geoff Kuenning."PARAID: A gear-shifting power-aware RAID".Journal ACM Transactions on Storage (TOS) TOS Homepage archive Volume 3 Issue 3, October 2007.
[7] Tseng-Yi Chen,Tsung-Tai Yeh,Hsin-Wen Wei,Yu-Xun Fang,Wei-Kuan Shih,Tsan-sheng Hsu."CacheRAID： An Effencient Adaptive Write Cache Policy to Conserve RAID Disk Array Energy".UCC,2012.
[8]. Facebook, Flashcache, https://github.com/facebook/flashcache/
[9]. Device Mapper, http://sourceware.org/dm/
[10] Tsung-Tai Yeh, Hsin-Wen Wei, Shin-Hao Liu, Pei-Chi Hung, Tsansheng Hsu, Yen-Chiu Chen, "The Development of Digital Archives Management Tools for iRODS", In Proceedings iRODS User Group Meeting, March, 2010
[11] Hsiu-Lien Yeh, Yan-Fu Chen, Tsung-Tai Yeh, Pei-Chi Huang, Shin- Hao Liu, Hsin-Wen Wei, and Tsan-sheng Hsu, "The Monitoring System Based on Nagios for Data Grid Environment." in International Conference on Grid Computing and Applications, July, 2011.
[12] http://iotta.snia.org/traces/414
[13] Joukov and Josef Sipek, "Greenfs: making enterprise computers greener by protecting them better," in EuroSys’08, Proceedings of the 3rd ACM SIGOPS/EuroSys Eruopean Conference on Computer Systems, 2008.
[14] Yuhui Deng. "What is the future of disk drives, death or rebirth?" In ACM Comput. Surv., March 2008.
[15] Segate corporation, "Hard drive datasheet," http://www.seagate.com/files/staticfiles/docs/pdf/datasheet/disc/barracuda-ds1737-1-1111us.pdf
[16] Jay J. Wylie and Ram Swaminathan Hewlett-Packard Labsjay.wyli. "Determining fault tolerance of XOR-based erasure codes efficiently." Published in proceedings of DSN 2007, The 37th Annual IEEE/IFIP International Conference on Dependable Systems and Networks, June 25 - June 28, 2007, Edinburgh, UK.
[17] Kevin M. Greenan, Xiaozhou Li,Jay J. Wylie. "Flat XOR-based erasure codes in storage systems:Constructions, efficient recovery, and tradeoffs." in Mass Storage Systems and Technologies (MSST), 2010.
[18] Kevin M. Greenan, "Reliability and Power-Efficiency inErasure-Coded Storage SystemsTechnical." in Technical Report UCSC-SSRC. 2008.
[19] Zhu, Q."Power-aware storage cache management".Computers, IEEE Transactions on May 2005.
[20] http://www.reliable-savings.com/
[21] Akshat Verma,Ricardo Koller,Luis Useche,Raju Rangaswami."SRCMap: Energy Proportional Storage using Dynamic Consolidation".