NetFPGA上基於OpenFlow的跨雲端動態資源調配機制之設計與實作

Design and Implementation of OpenFlow-based Dynamic Resource Allocation Mechanism in Inter-Cloud on NetFPGA

張德勤(Der-Chyn Chang)

指導教授 ： 周立德

國立中央大學/資訊電機學院/資訊工程學系/碩士(2012年)

摘要

近幾年雲端運算（Cloud Computing）受到企業或政府普遍的使用，也使得雲端中之資源管理成為一個重要的研究議題。雲端運算的概念是源自於分散式系統及網格運算，並且透過虛擬化技術，將實體機器上的硬體資源共享給不同使用者，充分使用實體資源，增加資源使用率，間接地降低企業成本。隨著舉世聞名的大公司帶動之下，建立在雲端運算上的雲端服務也越來越多，也越來越多國際公司提供雲端服務；但是目前雲與雲之間仍然欠缺一個有效的跨雲端之資源調配機制，使得雲與雲之間缺少一個互助性；而雲端間之互助能有效改善雲端服務的達成率以及使用者之感受程度，直接或間接地讓雲端提供者得到利益。本論文實作一個跨雲端之動態資源調配機制來改善現今跨雲間之互助性問題。
本論文提出一個基於模糊理論之動態資源調配機制（Fuzzy-based Dynamic Resource Allocation，簡稱FDRA），並且在各個雲端環境中實作一個雲端代理人，透過各個雲端上的雲端代理人統整各自的雲端資源之資訊，搭配本論文所提出的FDRA機制來配給雲端資源，提供一個動態地資源調配方式，以達到雲端中之資源使用負載平衡，以及跨雲端的資源互助。系統架構主要為裝載有NetFPGA網路開發板的雲端代理人及其所屬的雲端叢集，其中雲端叢集內部皆以OpenFlow Protocol互相溝通，基於此架構之上，建立本論文所提之FDRA機制，並且完成實作。最後，本論文所提機制與DAIRS機制做比較。實驗結果顯示，本論文所提之FDRA最多比DAIRS好上37.2%，顯示更能夠在資源負載不均衡的情形下調配資源改善負載不均衡之情形。

關鍵字

雲端運算 ； 資源調配 ； 跨雲端 ； OpenFlow ； NetFPGA ； 模糊理論 ； 資源互助

並列摘要

In recent years, Cloud Computing is popular with most enterprises and government and also makes the cloud resource management has become an important research topics. The concept of cloud computing follows distributed systems and grid computing, and shares the physical resources to different users through the virtualization technologies in order to increase the resource utilization and reduce the costs. With the leading of world-famous companies, the number of cloud services is increased rapidly, but an effective resource allocation mechanism is still scarce between each cloud, which makes the lack of a resource mutual aid in inter-cloud. Nevertheless, inter-cloud with such assistance can improve the achievement rates of cloud services and the degree of users’ feelings, bringing the benefits to the service provider directly or indirectly. In this paper, we proposed a dynamic resource allocation mechanism to improve the insufficient of mutual assistance nowadays in inter-cloud.
This paper presents a Fuzzy-based Dynamic Resource Allocation mechanism (FDRA), and a Cloud Broker is implemented in each cloud. The Cloud Broker not only integrates the information of each cloud cluster, but also reallocates the cloud resources with FDRA to achieve the load balance and the resource mutual assistance. The system architecture includes NetFPGA-based Cloud Brokers and Cloud Broker-owned cloud clusters. The OpenFlow protocol is deployed to provide the communication between the computing nodes and a Cloud Broker in each cloud cluster. Finally, the proposed FDRA is compared with DAIRS mechanism. The experimental results show that FDRA is 37.2% times better than DAIRS in best case, especially under the condition of the load imbalance of cloud resource.

並列關鍵字

OpenFlow ； NetFPGA ； Cloud Computing ； Inter-Cloud ； Resource Allocation ； Resource Mutual Assistance ； Fuzzy

參考文獻

[5] W. Tian, S. Su and G. Lu, “A Framework for Implementing and Managing Platform as a Service in a Virtual Cloud Computing Lab,” Proceedings of 2010 the Second International Workshop on Education Technology and Computer Science, Wuhan, China, pp. 273-276, 6-7, Mar. 2010.

[13] Eucalyptus, http://open.eucalyptus.com/

[26] J.-T. Piao and J. Yan, “A Network-aware Virtual Machine Placement and Migration Approach in Cloud Computing,” Proceedings of International Conference on Grid and Cloud Computing, Nanjing, China, pp. 87-92, 1-5, Nov. 2010.

[27] Y. Ke, G. Jianhua, Z. Tianhai and S. Guofei, “An Optimized Control Strategy for Load Balancing based on Live Migration of Virtual Machine,” Proceedings of Annual ChinaGrid Conference, Dalian, China, pp. 141-146, 22-23, Aug. 2011.

[30] A.-M. Alakeel, “A Fuzzy Dynamic Load Balancing Algorithm for Homogenous Distributed Systems,” World Academy of Science, Engineering and Technology, Issue 61, pp. 30-33, Jan. 2012.