雲端存儲服務如亞馬遜簡易儲存服務因簡化架構設計以及降低維護成本的特性，在近年來於使用者以及服務提供商中越來越受歡迎。但因其與傳統POSIX介面存儲系統架構上的不同，使用者並無法直接於其上運行應用程式。因此，越來越多研究開始致力開發探討將雲端存儲及本地POSIX存儲連結，希望能同時利用兩者帶來的好處。
本研究討論了支援部分修改(partial modification)及隨機存取(random access)等POSIX標準指令於雲端存儲上實作時所遭遇的挑戰與困難。雖然時至今日已有一些研究提出了使用固定大小分隔存放的方式來提升系統效能，但在實務上，需要更嚴謹地決定分隔大小。否則將因分隔過大增加大量無用資料傳輸量，或分隔過小觸發過多雲存取請求，降低整體系統效能。
此外，在先前研究中，我們已提出理論系統架構以及能夠線上分析I/O分布範式的啟發式策略，以動態地調整雲端資料分隔大小與分布來達到更好的效能。在此研究中，我們將談討啟發式策略的限制，並介紹模型分析策略以更精確的數學模型來描述及調整雲資料分布。最後，以實驗證明我們的策略確實能夠動態調整雲端資料分隔分布，且與固定大小分布策略相比增進7% ~ 106%的效能。

Cloud storage services, like Amazon S3, has become more and more popular for users and enterprises due to its simplified architecture design and reduced maintenance cost.
However, due to architecture differences between POSIX-interfaced file systems and cloud storage systems,
users can not transparently run most applications directly on a cloud storage system.
Therefore, more and more researchers are interested in cloud-backed file systems aiming to deliver a system that leverage the best of both POSIX-interface and cloud storage.

In this thesis, we discuss the challenges of supporting partial modification and random access of POSIX-interface operations on the cloud storage.
Existing frameworks have proposed fixed-sized partitioning strategies on a file to improve the performance.
However, deciding the size of these split files, so called blocks, is not trivial.
In more detailed, too large blocks may contain more irrelevant data for serving an I/O request which introduces more transferring overhead.
On the contrary, too small blocks may trigger more cloud requests to serve an I/O request which lowers the network utilization in the latency way.

In our previous work, we have proposed a theoretical system architecture and a heuristic method to dynamically capture the I/O access pattern of users' requests and adjust file blocks on the cloud to achieve better performance.
In this thesis, we show the limitation of the heuristic method, further improve the method by a mathematical model and introduced a model-based partitioning strategy.
We show that our strategies can analysis I/O patterns and adjust the blocks on the cloud to achieve better performance.
Our evaluation shows that our strategies can improve the overall performance by 7\% to 106\% compared to the static fixed-sized partitioning strategy.

1 Introduction 5
2 Motivation 8
2.1 Timing Breakdown of a Record 8
2.2 Transfer-Latency Trade-off 9
3 System Architecture 11
3.1 Cloud Storage 11
3.2 Meta-data Server 11
3.3 Client 12
3.4 Daemon 13
4 Heuristic Strategy 14
4.1 The Concept of Heuristic Method 14
4.2 Tracking Internal Byte Usage 14
4.3 Decision of merge-is-better and split-is-better 16
4.4 Rearrange Blocks List for Daemon 17
5 Model Based Strategy 18
5.1 The Concept of Region 18
5.2 Classifying Regions 20
5.2.1 Stability of a Region 20
5.2.2 Aligness of a Region 21
5.2.3 Unstable Region 23
5.3 Optimal Solution of a Region 24
5.3.1 Stable Aligned Region 24
5.3.2 Stable Misaligned Region 26
6 Evaluation 28
6.1 Environment and Parameters Settings 28
6.2 Workloads 29
6.3 Overall Performance 31
6.4 Block Size Distribution 33
6.5 Adaptiveness Over Time 34
6.6 Overhead 36
6.6.1 Blocks Locking Overhead 36
6.6.2 Analysis Overhead 36
6.6.3 Offset Algorithm Overhead 37
7 Related Works 39
8 Conclusion 42

[1] Amazon Simple Storage Service(S3)." http://aws.amazon.com/s3.
[2] Dropbox." https://www.dropbox.com.
[3] S3FS - FUSE-based file system backed by Amazon S3." https://github.com/
s3fs-fuse/s3fs-fuse.
[4] S3QL - a full-featured file system for online data storage." https://bitbucket.
org/nikratio/s3ql/.
[5] M. Vrable, S. Savage, and G. M. Voelker, \Bluesky: A Cloud-Backed File System
for the Enterprise," 10th USENIX Conference on File and Storage Tech-
nologies (FAST), 2012.
[6] A. Bessani, R. Mendes, T. Oliveira, N. Neves, M. Correia, M. Pasin, and
P. Verissimo, SCFS: A Shared Cloud-backed File System," USENIX Annual
Technical Conference (USENIX), 2014.
[7] Y. Abe and G. Gibson, pWalrus: Towards Better Integration of Parallel File
Systems into Cloud Storage," Workshop on Interfaces and Abstraction for Sci-
entific Data Storage (IASDS), co-located with IEEE Int. Conference on Cluster
Computing 2010 (Cluster), 2010.
[8] Y. Kuo, Y. Jeng, and J. Chen, A Hybrid Cloud Storage Architecture for
Service Operational High Avalibility," IEEE 37th Annual Computer Software
and Applications Conference (COMPSAC), 2013.
[9] ibench traces," http://research.cs.wisc.edu/adsl/Traces/ibench/.
43
[10] A. D. D. Narayanan and M. R. L. A. Rowstron, Write Off-Loading: Practical
Power Management for Enterprise Storage," 6th USENIX Conference on File
and Storage Technologies (FAST), 2008.
[11] K. P. N. Puttaswamy, T. Nandagopal, and M. Kodialam, Frugal Storage for
Cloud File Systems," Proceedings of the 7th ACM european conference on Com-
puter Systems (EuroSys), 2012.
[12] Amazon Elastic Block Store." https://aws.amazon.com/ebs/.
[13] Amazon ElastiCache." https://aws.amazon.com/elasticache.
[14] H. Abu-Libdeh, L. Princehouse, and H. Weatherspoon, RACS: A Case for
Cloud Storage Diversity," ACM Symposium on Cloud Computing (SoCC), 2010.
[15] A. Bessani, M. Correia, B. Quaresma, F. Andre, and P. Sousa, DEPSKY:
Dependable and Secure Storage in a Cloud-of-Clouds," ACM Transactions on
Storage (TOS), 2013.
[16] C. Cachin, R. Haas, and M. Vukolic, Dependable Storage in the Intercloud,"
IBM Research, 2010.
[17] D. Bermbach, M. Klems, S. Tai, and M. Menzel, Metastorage: A Federated
Cloud Storage System to Manage Consistency-Latency Tradeoffs," IEEE 4th
International Conference on Cloud Computing (CLOUD), 2011.
[18] G. DeCandia, D. Hastorun, M. Jampani, G. Kakulapati, A. Lakshman,
A. Pilchin, S. Sivasubramanian, P. Vosshall, and W. Vogels, Dynamo: Amazons
Highly Available Key-value Store," 21st ACM SIGOPS symposium on
Operating systems principles (SOSP), 2007.
[19] Y. Hu, H. C. H. Chen, P. P. C. Lee, and Y. Tang, NCCloud: Applying Network
Coding for the Storage Repair in a Cloud-of-Clouds," 10th USENIX Conference
on File and Storage Technologies (FAST), 2011.
[20] X. Ding, S. Jiang, F. Chen, K. Davis, and X. Zhang, Diskseen: Exploiting
Disk Layout and Access History to Enhance I/O Prefetch," USENIX Annual
Technical Conference (USENIX), 2007.