近年來，雲端服務改變以往傳統網路服務的使用型態，雲端服務提供服務平台用於建立網路上的應用服務並由提供商來維護後端的服務主機，讓使用者不在需要額外花費成本與心力去維護服務的硬體平台，其中雲端服務所能發展的原因，虛擬化技術扮演了很重要的角色，虛擬化技術提供了在主機上同時運行多個虛擬系統的原理，讓這些虛擬系統可一同使用主機上的硬體資源，使硬體資源可以有效率的被使用，雲端服務可以透過租貸的方式提供這些虛擬系統給使用者進行使用，從而提高服務主機的收益並使得雲端服務得以發展。

然而在虛擬化技術中，虛擬系統都是同時共享著主機的硬體資源，也因此使得硬體使用效率會受到這些虛擬系統的運行受到干擾從而讓處理效率會有所降低，其中以硬碟I/O處理最為影響。而為了解決上述議題，有相關的研究提出了保證I/O處理量與滿足處理時間的排程，以改善系統對硬體的使用干擾性，但除了在虛擬化技術領域有所研究外，在多媒體領域上也有硬碟I/O處理的研究存在著，同時多媒體領域提出的硬碟I/O排程的概念與虛擬化技術提出的硬碟I/O排程有其相似存在，而這是否也意謂著前者的硬碟I/O排程也可以適用於虛擬化技術上使用。

因此，在本篇論文中，我們採用多媒體領域中的DM-SCAN排程實作在Xen虛擬平台中進行實驗，而實驗結果可得DM-SCAN排程在sequential讀取與random讀取測試中可以比EDF提高約5.1%與15.1%的提升，同時在提升硬碟I/O處理的情況下，也能比Deadline與Flubber改善近8成的miss deadline情況，另一方面，我們也在Xen中實作可用於控制虛擬系統的硬碟I/O的處理架構，並在實驗中可得該架構可以成功的控制虛擬系統的硬碟I/O處理量，完成在Xen可進行服務品質的機制。

Cloud Service changes the traditional service of internet in recent years. Cloud Service Provider provides user with service system to build the service. User does not care what hardware happened in the system, and only cares about the mantenance of service. The reason that cloud service can support user to build service is virtualization technology. It provides cloud service provider to create and manage multiple virtual system in one computer, and each system is isolated in computer. In this techology, Hardware can have the high Utilization in computer and also increase earnings for cloud service provider.

Virtualization technology improves the utilization of hardware in computer. However, it also cause the interference for each virtual system because virtual systems share the physical hardware. The interference can reduce the performance of hardware when all virtual system is high loading,ex: disk. To reduce the interference of disk, some researches provides the solution to improve throughput and latency for disk and virtual system. On the other hand, there are also some solution to improve throughput and latency for disk in multimedia systems. To compare with solutions in virtualization technology, some idea are similar for improving the throughput with disk. It also means whether or not solution in multimedia can be used in virtualization technology.

In this paper, we implement DM-SCAN scheduling for multimedia in Xen virtualization platform. We compare the throughput and latency with EDF, Deadline, and Flubber. DM-SCAN can improve 5.1% and 15.1% throughput in sequential read and random read which compared with EDF. Therefore, In low deadline setting, DM-SCAN can improve the amount of miss deadline compared with Deadline and Flubber. It means DM-SCAN can be used in Xen and has the better improvement in disk interference.

1.介紹
2.研究背景
2.1.虛擬化技術介紹
2.2.Xen虛擬平台介紹
2.3.虛擬技術中硬碟I/O排程研究的分析
2.4.虛擬技術中硬碟I/O處理行為的分析
3.Xen平台中DM-SCAN實現流程
3.1.架構說明
3.2.各功能實現介紹
3.3.服務品質控制機制實現
4.實驗結果
4.1.實驗環境介紹
4.2.效能測試工具介紹
4.3.實驗結果與分析
4.3.1 每秒I/O處理量與miss deadline結果
4.3.2 I/O要求的執行時間統計結果
4.3.3 硬碟I/O處理控制機制驗證結果
5.結論與未來研究方向
6.引用

[1]. Ray-I Chang, Wei-Kuan Shih, and Ruei-Chuan Chang. Real-Time Disk Scheduling for Multimedia Applications with Deadline-Modification-Scan Scheme. Real-Time Systems, Volume 19 Issue 2, Page 149-168, 2000.
[2]. Paul Barham, Boris Dragovic, et al. Xen and the Art of Virtualization. Proceedings of the nineteenth ACM symposium on Operating systems principles, Volume 37 Issue 5, Pages 164-177, ACM, 2003.
[3]. Hypervisor(online), wikipedia, 2014. from http://en.wikipedia.org/wiki/Hypervisor .
[4]. Virtual Machine Manager(online), wikipedia, 2013. From
http://en.wikipedia.org/wiki/Virtual_Machine_Manager .
[5]. Ring(online), wikipedia, 2014. from
http://en.wikipedia.org/wiki/Ring_(computer_security) .
[6]. Avi Kivity, Yaniv Kamay, et al. kvm: the Linux Virtual Machine Monitor. Linux Symposium, Volume 1, Page 225-230, 2007.
[7]. Fabrice Bellard. QEMU, a Fast and Portable Dynamic Translator. Proceedings of the annual conference on USENIX Annual Technical Conference, USENIX, Page 41-46, 2005.
[8]. VMware white paper(online), VMware. from
https://www.vmware.com/pdf/virtualization.pdf .
[9]. Platform virtualization(online), wikipedia, 2014. from
http://en.wikipedia.org/wiki/Platform_virtualization .
[10]. Hyper-V(online), wikipedia, 2014. from http://en.wikipedia.org/wiki/Hyper-V .
[11]. XenServer(online), citrix, 1999. from http://www.citrix.com/ .
[12]. Xen(online), wikipedia, 2014. from http://en.wikipedia.org/wiki/Xen .
[13]. Rusty Russell. virtio: Towards a De-Facto Standard For Virtual I/O Devices. ACM SIGOPS Operating Systems Review - Research and developments in the Linux kernel, Volume 42 Issue 5, Page 95-103, ACM, 2008.
[14]. Xen Blktap(online), Xen wikipedia, 2013. from http://wiki.xen.org/wiki/Blktap .
[15]. Hwanju Kim, Hyeontaek Lim, Jinkyu Jeong, et al. Task-aware virtual machine scheduling for I/O performance. Proceedings of the 2009 ACM SIGPLAN/SIGOPS international conference on Virtual execution environments, Pages 101-110, ACM, 2009.
[16]. Xiaolin Wang, Jiarui Zang, et al. Selective hardware/software memory virtualization. Proceedings of the 7th ACM SIGPLAN/SIGOPS international conference on Virtual execution environments, Volume 46 Issue 7, Page 217-226, ACM, 2011.
[17]. Aravind Menon, Alan L. Cox, and Willy Zwaenepoel. Optimizing network virtualization in Xen. Proceedings of the 2006 USENIX Annual Technical Conference, Pages 2-2, USENIX, 2006.
[18]. Michael Myers and Stephen Youndt. An Introduction to Hardware-Assisted Virtual Machine (HVM) Rootkits. Crucial Security, 2007.
[19]. Linux I/O Scheduler(online), LinuxQuestions, 2007. from
http://wiki.linuxquestions.org/wiki/IOSched .
[20]. Vladimir Stantchev and Christian Schröpfer. Negotiating and Enforcing QoS and SLAs in Grid and Cloud Computing. Advances in Grid and Pervasive Computing, Volume 5529, Pages 25-35, 2009.
[21]. Shadi Ibrahim, Bingsheng He, and Hai Jin. Towards Pay-As-You-Consume Cloud Computing. IEEE International Conference on Services Computing, Pages 370-377, IEEE, 2011.
[22]. Seetharami R. Seelam and Patricia J. Teller. Virtual I/O Scheduler: A Scheduler of Schedulers for Performance Virtualization. Proceedings of the 3rd international conference on Virtual execution environments, Pages 105-115, ACM, 2007.
[23]. Ajay Gulati, Irfan Ahmad, and Carl A. Waldspurger. PARDA: Proportional Allocation of Resources for Distributed Storage Access. Proccedings of the 7th conference on File and storage technologies, Pages 85-98, USENIX, 2009.
[24]. Ajay Gulati, Arif Merchant, and Peter J. Varman. pClock: An Arrival Curve Based Approach For QoS Guarantees In Shared Storage Systems. Proceedings of the 2007 ACM SIGMETRICS international conference on Measurement and modeling of computer systems, Volume 35 Issue 1, Pages 13-24, ACM, 2007.
[25]. Jianyong Zhang, Anand Sivasubramaniam, et al. Storage Performance Virtualization via Throughput and Latency Control. ACM Transactions on Storage, Volume 2 Issue 3, Pages 283-308, ACM, 2006.
[26]. Hai Jin, Xiao Ling, Shadi Ibrahim, et al. Flubber: Two-level Disk Scheduling in Virtualized Environment. Future Generation Computer Systems, Volume 29 Issue 8, Pages 2222-2238, 2013.
[27]. Ajay Gulati, Arif Merchant, Mustafa Uysal, et al. Efficient and adaptive proportional share I/O scheduling. Technical Report HPL-2007-186, HP Labs, 2007.
[28]. P. Valente and F. Checconi. High throughput disk scheduling with fair bandwidth distribution. IEEE Transactions on Computers, Volume 59 Issue 9, Pages 1172-1176, IEEE, 2010.
[29]. A. L. N. Reddy , Jim Wyllie , and K. B. R. Wijayaratne. Disk Scheduling in a Multimedia I/O system. Proceedings of ACM Multimedia, Volume 1 Issue 1, Pages 37-59, ACM, 1993.
[30]. Kartik Gopalan. Real-Time Disk Scheduling Using Deadline Sensitive SCAN. State University of New York at Stony Brook, 2001.
[31]. Seagate Disk Spec(online), Seagate, 2008. from
http://www.seagate.com/staticfiles/support/disc/manuals/desktop/Barracuda%207200.11/100507013e.pdf .
[32]. Alexey Kopytov. Sysbench(online), 2004. from https://launchpad.net/sysbench .