雲段儲存服務的使用者日益增多，資源開始不足的議題是無可避免的。而在這個時候，雲端儲存服務的水平擴充的架構就能夠解決資源不足的問題。當我們部署服務在單一節點上時，資源很明顯是不夠的。很多現有的雲端儲存服務，像是Dropbox、Google Drive和Amazon S3也使用叢集的方式來部署服務以提供更多的資源，因此把水平擴充的架構部署到叢集上，對雲端儲存服務來說是個其中一個重要的趨勢。
在這篇論文裡我們呈現一個基於容器技術的水平擴充架構，能解決雲端儲存服務資源不足的問題。我們基於這樣的架構實作了SSBox叢集系統。基於容器技術的SSBox叢集系統是基於Docker和CoreOS實作的。Docker利用container技術讓部署方便，而CoreOS則是使用容器技術的orchestrator。我們的目標是藉由使用這些這些技術，水平擴充雲端儲存服務並取得效能的提升跟提供動態部署管理的功能。實驗的結果顯示我們的架構可以線性的提升效能，而架構因為很彈性，也可以被其他雲端服務使用。

The concept and implement of the scale-out architecture for cloud storage service is significant and necessary due to the increased number of users. The resources in a single node is definitely not enough. Many famous cloud storage services like Dropbox, Google Drive and Amazon S3 also build their services on a cluster to provide more resources. Therefore, the scale-out architecture on a cluster is one of the most important trends for cloud storage services.
In this thesis, we propose our container-based scale-out architecture for cloud storage service. Based on this architecture, we implement SSBox cluster system. The container-based SSBox cluster system is based on Docker and CoreOS. Docker is the advanced container technology for deployment, and CoreOS is the orchestrator for containers. Our goal is to scale out cloud storage service to get performance boost and provide dynamic deployment management with these technologies. The experimental result shows that SSBox scales out linearly based on the proposed architecture. Since the architecture is flexible, it can also be applied to other cloud services.

List of Figures 4
List of Tables 6
Chapter 1. Introduction 7
Chapter 2. Related Work 9
2.1 Cloud Storage Service 9
2.2 Scale-Out Architecture 9
2.3 Container-Based Deployment 10
Chapter 3. Container-Based SSBox Cluster System 12
3.1 Design Methodology 12
3.2 Single Node Architecture 13
3.2.1 SSBox APIs 14
3.3 Scale-Out Architecture 18
3.4 Container-Based Scale-Out Architecture 19
Chapter 4. System Optimization 25
4.1 Docker Container 25
4.2 Nginx Optimization 27
4.2.1 Worker 27
4.2.2 Keep Alive 28
4.2.3 Send File 28
4.2.4 Access logs 29
4.2.5 Buffer 29
4.3 PostgreSQL Optimization 30
4.4 Scale-Out Factor 30
4.4.1 Thread 31
4.4.2 Process 31
4.4.3 Container 31
4.4.4 Node 32
4.5 Ceph File System 33
Chapter 5. Experiments 34
5.1 Preliminary 34
5.1.1 Google Compute Engine 34
5.1.2 loader.io 34
5.2 Experimental Environment 35
5.3 Single Node Optimization 35
5.3.1 Thread 36
5.3.2 Process 39
5.3.3 Combinations of Containers and Processes 42
5.4 Scale-Out of Multiple Nodes 43
5.5 Load Test 46
Chapter 6. Conclusions and Future Work 49
REFERENCE 50

1. Dropbox. Available from: https://http://www.dropbox.com/.
2. Google Drive. Available from: https://http://www.google.com/intl/zh-TW/drive/.
3. CoreOS. Available from: https://coreos.com/.
4. Docker. Available from: https://http://www.docker.com/ .
5. Google Compute Engine. Available from: https://cloud.google.com/compute/.
6. loader. Available from: https://loader.io/.
7. Amazon S3. Available from: http://aws.amazon.com/tw/s3/.
8. Dropbox (service) wiki. Available from: https://en.wikipedia.org/wiki/Dropbox_(service).
9. Eranki, R. Scaling lessons learned at Dropbox, part 1. 2012; Available from: http://eranki.tumblr.com/post/27076431887/scaling-lessons-learned-at-dropbox-part-1.
10. Google Drive wiki. Available from: https://en.wikipedia.org/wiki/Google_Drive.
11. ownCloud. Available from: https://owncloud.org/.
12. ownCloud wiki. Available from: https://en.wikipedia.org/wiki/OwnCloud.
13. Hildmann, T. and O. Kao. Deploying and extending on-premise cloud storage based on ownCloud. in Distributed Computing Systems Workshops (ICDCSW), 2014 IEEE 34th International Conference on. 2014. IEEE.
14. Weil, S.A., et al. Ceph: A scalable, high-performance distributed file system. in Proceedings of the 7th symposium on Operating systems design and implementation. 2006. USENIX Association.
15. Ceph. Available from: http://ceph.com/ .
16. Ghemawat, S., H. Gobioff, and S.-T. Leung. The Google file system. in ACM SIGOPS operating systems review. 2003. ACM.
17. DeCandia, G., et al. Dynamo: amazon's highly available key-value store. in ACM SIGOPS Operating Systems Review. 2007. ACM.
18. Ho, S.-S., Distributed Metaserver Mechanism and Recovery Mechanism Support in Quantcast File System, in Department of Institute of Information Systems and Applications. 2014, National Tsing Hua University. p. 48.
19. Vazquez, C., et al. Dynamic provision of computing resources from grid infrastructures and cloud providers. in Grid and Pervasive Computing Conference, 2009. GPC'09. Workshops at the. 2009. IEEE.
20. Deis. Available from: http://docs.deis.io/en/latest/.
21. Mell, P. and T. Grance, The NIST definition of cloud computing. 2011.
22. Chen, P.F. and N. Kodirov, Virtual Middlebox Management for Cloud. 2015.
23. Sekar, V., et al. Design and implementation of a consolidated middlebox architecture. in Proceedings of the 9th USENIX conference on Networked Systems Design and Implementation. 2012. USENIX Association.
24. Gember, A., et al., Stratos: A network-aware orchestration layer for middleboxes in the cloud. 2013, Technical Report.
25. Gerlach, W., et al. Skyport: container-based execution environment management for multi-cloud scientific workflows. in Proceedings of the 5th International Workshop on Data-Intensive Computing in the Clouds. 2014. IEEE Press.
26. Momjian, B., PostgreSQL: introduction and concepts. Vol. 192. 2001: Addison-Wesley New York.
27. PostgreSQL. Available from: http://www.postgresql.org/.
28. Donvito, G., G. Marzulli, and D. Diacono. Testing of several distributed file-systems (HDFS, Ceph and GlusterFS) for supporting the HEP experiments analysis. in Journal of Physics: Conference Series. 2014. IOP Publishing.
29. redhat. Linux contaienrs with docker format. Available from: https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/7/html/7.0_Release_Notes/chap-Red_Hat_Enterprise_Linux-7.0_Release_Notes-Linux_Containers_with_Docker_Format.html.
30. Felter, W., et al., An updated performance comparison of virtual machines and linux containers. technology, 2014. 28: p. 32.
31. Tuning Your PostgreSQL Server. Available from: https://wiki.postgresql.org/wiki/Tuning_Your_PostgreSQL_Server.
32. Kubernetes. Available from: http://kubernetes.io/.