云計(jì)算平臺(tái)虛擬機(jī)-資料下載頁
2025-06-23 14:57本頁面
【正文】 wall time for booting a VCN was substantially less than the time required for booting a physical pute node. Furthermore, the similar times for the actual Linux boot processes indicates that KVM disk I/O overhead is negligable in the context of basic system operations. Thus, performing maintenance tasks that require rebooting the system should result in less downtime for a Virtual Organization Cluster when pared to an equivalent physical cluster. As shown in Table III, KVM was ecient when network I/O is not a factor. Unfortunately, poor network performance was observed, as shown by the large (7578% overheads present in MPI jobs. Additional testing will be needed to determine the cause of this performance loss and resolve the underlying issue. Poor network performance could be attributed to several factors, which will require more testing to dierentiate: Since the Spanning Tree Protocol (STP was not enabled on our TUN/TAP bridges, routing loops might have been present, introducing a large amount of network latency Each VCN on a physical node had a 1Gbps link to the bridge, but each physical node only has 1Gbps to the switch, collisions might have occurred. Such collisions could have caused binaryexponential backo, thus crippling latency and bandwidth under heavy load. The code for the emulated Gigabit Ethernet NIC in QEMU was a recent addition at the time of this experiment, and it could have been inplete. ACM Transactions on Computational Logic, Vol. V, No. N, Month 20YY. An Evaluation of KVM for Use in Cloud Computing 7 The loss in performance of networksensitive jobs represents only a small fraction of the total number of scientic jobs. This makes our model particularly well These jobs are suited to Condor jobs in the vanilla and standard universes. and return their results. putationally expensive, but only rely on network I/O to get their initial state So, even though Condor has MPI capability, it is not really a limiting factor because most sites only provide support for the vanilla and standard universes [Thain et al. 2003]. The Virtual Organization Cluster Model provides great potential for cloud puting, since it denes a minimum specication for VOC support. A VO is not tied to a particular site for its puting resources but is instead free to move between sites that support the VOCM. This migration could be as simple as bringing down the VOC, copying two disk images and a small conguration le, and booting the VOC on a pletely dierent site. REFERENCES Habib, I. IBM. 2008. Virtualization with kvm. Linux Journal 2008, 166 (February, 8. 2005. Ibm systems virtualization. Jones, M. T. 2006. Virtual linux. Tech. rep., IBM. December. Keahey, K., Doering, K., and Foster, I. 2004. From sandbox to playground: Dynamic virtual environments in the grid. In 5th International Workshop on Grid Computing (Grid 2004. Qumranet. 2006. Kvm kernelbased virtualiztion machine white paper. Thain, D., Tannenbaum, T., and Livny, M. 2003. Condor and the Grid. Wiley, Chapter 11, 299350. van Doorn, L. 2006. Hardware virtualization trends. In Second International Conference on Virtual Execution Environments. ACM Transactions on Computational Logic, Vol. V, No. N, Month 20YY. 寧可累死在路上，也不能閑死在家里！寧可去碰壁，也不能面壁。是狼就要練好牙，是羊就要練好腿。什么是奮斗？奮斗就是每天很難，可一年一年卻越來越容易。不奮斗就是每天都很容易，可一年一年越來越難。能干的人，不在情緒上計(jì)較，只在做事上認(rèn)真；無能的人！不在做事上認(rèn)真，只在情緒上計(jì)較。拼一個(gè)春夏秋冬！贏一個(gè)無悔人生！早安！—————獻(xiàn)給所有努力的人.學(xué)習(xí)
研究報(bào)告相關(guān)推薦
鄂ICP備17016276號(hào)-1