近年來雲端儲存系統隨著網路與行動網路的普及越來越被廣泛使用，但隨之資料安全性的問題也越來越成為使用者的顧慮。不同於以往，許多較新的專案為了方便統整管理雲端儲存或解決安全性的問題，將多個雲端儲存結合成分散式雲端儲存系統。分散式雲端儲存系統雖然擁有較高的安全性，但目前的專案或產品都是以檔案為單位，並沒有實際運用到多個雲端儲存所帶來的效益。要在分散式雲端儲存上達到效能最佳化，我們提出了一個基於頻寬與容量的方法，透過資料的上傳與更新將常用的資料塊做頻寬使用優化，並在資料搬移時將不常用的資料塊做容量平衡；對此，我們提出了窗口為基準的常用資料塊辨識方法，並提出了個計數器的簡化版與另一個再簡化版來提高效率。最後，我們設計了一個實驗來模擬我們的方法，利用真實的測試資料，並與現行常見的方法做比較，結果顯示我們的發法在效能上優秀許多。藉由我們所提出的架構不只能解決資料安全性的問題，也達到效能最佳化的目的。

Due to the elasticity and convenience of usage, cloud storage systems have played an important role in the market of data storage. However, as user data are stored on cloud storage systems hosted by cloud service providers, their privacy against unauthorized accesses becomes a critical concern. In the meantime, it is an interesting and important issue on how to consolidate multiple heterogeneous cloud storage systems as one virtual storage space, so as to fully utilize the storage space and access bandwidth of the cloud storage systems. There are still some missing parts in the joint optimization of storage space and access bandwidth for a high-performance, economical virtual storage system. This work proposes a novel scheme for the storage arbiter, which performs the storage consolidation and various data management tasks, such as the data placement, migration, and indexing. The storage arbiter could be hosted on a private server or a secure cloud service, so as to ensure the privacy of user data. In addition, we present a window-based technique for hot data identification to support the data management tasks, such as data migration. To evaluate the presented scheme and technique, we used trace-driven simulation on real-world workloads, where the experimental results are very encouraging.

1.Introduction 1
2.System Architecture and Motivations 5
3.Joint Management Scheme with Storage Capacity and Bandwidth Considerations 11
3.1 Overview 11
3.2 Data Allocation & Migration with Storage Capacity and Bandwidth Considerations 13
3.3 Management of File and Chunk Mapping in the Metadata Server 18
3.4 Hot Chunk Identification Strategies 21
3.4.1 Window-based Hot Chunk Identification 21
3.4.2 Counter-based Hot Chunk Identification 24
3.4.3 Flag-based Hot Chunk Identification 24
4.Experimental Studies 26
4.1 Experimental Setup 26
4.2 Size Sensitivity 29
4.2.1 Total Size Sensitivity 30
4.3 Bandwidth Sensitivity 32
4.4 Stability 34
5.Conclusion 37
6.References 38

[1] Google, "Google Drive," [Online]. Available: https://drive.google.com. [Accessed 9th July 2015].
[2] Microsoft, "OneDrive," [Online]. Available: https://onedrive.live.com. [Accessed 9th July 2015].
[3] Box, "Box," [Online]. Available: https://www.box.com. [Accessed 9th Jult 2015].
[4] Dropbox, "Dropbox," [Online]. Available: https://www.dropbox.com. [Accessed 9th July 2015].
[5] Bplans, "Why cloud storage is growing in use and popularity," [Online]. Available: http://articles.bplans.com/why-cloud-storage-is-growing-in-use-and-popularity/. [Accessed 9th July 2015].
[6] unwire.hk, "Speed and size comparison of cloud storages (webpage in Chinese)," [Online]. Available: http://unwire.hk/2014/04/28/online_storage/top-_news/. [Accessed 9th July 2015].
[7] L. Huang, G. Peng and T.-c. Chiueh, "Multi-dimensional storage virtualization," Proceedings of the joint international conference on Measurement and modeling of computer systems, p. 14–24, June 2004.
[8] A. Singh, M. Korupolu and D. Mohapatra, "Server-storage virtualization: Integration and load balancing in data centers," International Conference for High Performance Computing, Networking, Storage and Analysis, November 2008.
[9] OneBigDrive, "OneBigDrive," [Online]. Available: https://onebigdrive.com.. [Accessed 9th July 2015].
[10] J. Strickland, "Concerns about Cloud Storage," [Online]. Available: http://computer.howstuffworks.com/cloud-computing/cloud-storage3.htm. [Accessed 9th July 2015].
[11] Cloudfuze, "Cloudfuze," [Online]. Available: https://www.cloudfuze.com. [Accessed 9th July 2015].
[12] MultCloud, "MultCloud.," [Online]. Available: https://www.multcloud.com. [Accessed 9th July 2015].
[13] Y. Gu and R. L. Grossman, "Sector: A high performance wide area community data storage and sharing system," Future Generation Computer Systems, pp. 720-728, May 2010.
[14] I. Repository, "Storage Networking Industry Association," [Online]. Available: http://iotta.snia.org/.
[15] S. Kavalanekar, B. Worthington, Q. Zhang and V. Sharda, "Characterization of storage workload traces from production Windows Servers," IEEE International Symposium on Workload Characterization, p. 119–128, September 2008.
[16] Srinivasan, J. ; Wei Wei ; Xiaosong Ma ; Ting Yu, "EMFS: Email-based Personal Cloud Storage," in 6th IEEE International Conference on Networking, Architecture and Storage (NAS), Dalian, Liaoning, 2011.