臺灣博碩士論文加值系統

現代小型伺服器與虛擬化技術的普及化，小型私有雲已成為個人或小企業提供網路服務或平行計算資源的有力選擇。小型私有雲受限於有限的硬體資源，與大型雲端環境相比，除了每台伺服器的硬體資源不一致，其不規則的網路拓樸也是一個鮮明的特色。
在虛擬機器的資源管理上，已有許多論文提出基於單台機器的資源監控與分配系統，或基於大型雲端的資源平衡與虛擬機器搬遷，但對於小型伺服器所構成小規模多主機的私有雲環境還沒有太多研究。
本篇研究中我們實作了一個虛擬化伺服器管理系統，透過垂直分析監控來評估虛擬機器中的服務品質，並以此資訊來進行更細緻的資源分配，藉以達到高效能或節能的目的。

With the popularity of microserver and virtualization technology, private cloud has already become a good choice to provide network service or parallel computing power. Compare to public cloud enviroment, ununified hardware resource and irregular network topology is the feature of privte cloud.
For virtualized resource management, previous works are mostly focusing on ether the hardware resource monitor and provision based on single host or resource balancing and virtual machine(VM) migration on public cloud, private cloud enviroment is seldom mentioned.
In this paper, we implement a virtualized server management system which uses vertical profiling monitor to estimate the Quality of service (QoS) of VM then uses this information providing fine-grained resource provision, finally achieve high performance or power saving management.

摘要　　　　　　 i
ABSTRACT ii
Table of Contents iv
List of Tables vi
List of Figures vii
I. Introduction 1
1.1 Background 1
1.2 Motivation 2
1.3 Contribution 4
1.4 Report Organization 5
II. Related Work 6
2.1 Page table entry memory monitor 6
2.2 Single host memory monitor compare 7
2.3 Multi-host VM 7
III. Methodology 9
3.1 Private cloud manager framework 9
3.2 System design 11
3.2.1 System environment 11
3.2.2 Resource controller 12
3.2.3 Vertical profiling monitor 13
3.2.4 Network monitor 15
3.2.5 Monitor target 16
3.2.6 Benchmarks 17
IV. Observation and provision policy 19
4.1 Baseline resource detection 19
4.2 Workload type 20
4.2.1 Fork workload 20
4.2.2 Dynamic workload 21
4.3 Resource provision policy 23
4.3.1 Memory demand prediction 23
4.3.2 VM distributing advice for high performance 23
4.3.3 VM distributing advice for power saving 24
V. Evaluation 25
5.1 Experiment Configuration 25
5.2 Result 26
VI. Conclusion 30
References 31

[1] Yan Niu, Chun Yang, Xu Cheng, “Dynamic Memory Demand Estimating Based on the Guest Operating System Behaviors for VMs,”, in Parallel and Distributed Processing with Applications (ISPA), 2011 IEEE 9th International Symposium on, pp 81 – 86 (2011)
[2] Weiming Zhao, Zhenlin Wang, “Dynamic memory balancing for VMs”, in VEE '09 Proceedings of the 2009 ACM SIGPLAN/SIGOPS international conference on Virtual execution environments, pp 21 – 30(2009)
[3] Lanzheng Liu, Rui Chu, Yongchun Zhu, Pengfei Zhang, Liufeng Wang, “DMSS: A Dynamic Memory Scheduling System in Server Consolidation Environments”, in Object/Component/Service-Oriented Real-Time Distributed Computing Workshops (ISORCW), 2012 15th IEEE International Symposium on, pp 70 – 75(2012)
[4] Diwaker Gupta, Sangmin Lee, Michael Vrable, Stefan Savage, Alex C. Snoeren, George Varghese, Geoffrey M. Voelker, Amin Vahdat, “Difference engine: harnessing memory redundancy in VMs” , in Communications of the ACM CACM Homepage archive Volume 53 Issue 10, October 2010, pp 85 – 93(2010)
[5] Heo, J., Xiaoyun Zhu, Padala, P., Zhikui Wang, “Memory overbooking and dynamic control of Xen VMs in consolidated environments”, in Integrated Network Management, 2009. IM '09. IFIP/IEEE International Symposium on, pp 630 – 637(2009)
[6] Abel Gordon, Michael Hines, Dilma Da Silva, Muli Ben-Yehuda, Marcio Silva, Gabriel Lizarraga, “Ginkgo: Automated, Application-Driven Memory Overcommitment for Cloud Computing”, In ASPLOS RESoLVE '11: Runtime Environments/Systems, Layering, and Virtualized Environments (RESoLVE) workshop (2011)