1、整理課件,1,談高校云計(jì)算的必要性和可行性,肖 波 中央民族大學(xué),整理課件,2,一、云計(jì)算是IT發(fā)展的趨勢,1980s 計(jì)算機(jī)就是計(jì)算機(jī)； 1990s 網(wǎng)絡(luò)就是計(jì)算機(jī) 2000s 數(shù)據(jù)中心就是計(jì)算機(jī) 2010- 云計(jì)算就是計(jì)算機(jī),整理課件,3,整理課件,4,整理課件,5,2011 Gartner “IT Hype Cycle” for Emerging Technologies,Copyright 2012, Elsevier Inc. All rights reserved,整理課件,6,Clouds and Internet of Things,HPC: High-Performance

2、Computing HTC: High-Throughput Computing P2P: Peer to Peer MPP: Massively Parallel Processors,整理課件,7,The Internet of Things (IoT,Smart Earth: An IBM Dream,整理課件,8,Cloud,云計(jì)算將IT相關(guān)的能力以服務(wù)的方式提供給用戶，允許用戶在不了解提供服務(wù)的技術(shù)、沒有相關(guān)知識以及設(shè)備操作能力的情況下，通過Internet獲取所需要的服務(wù),整理課件,9,Vaquerolm的定義 云是一個(gè)包含大量可用的虛擬資源（硬件、開發(fā)平臺、I/O服務(wù)）的資源池。

3、這些資源可以根據(jù)不同的負(fù)載動(dòng)態(tài)地重新配置，以達(dá)到更優(yōu)化的資源利用率,整理課件,10,The Next Revolution in ITCloud Computing,Classical Computing Buy & Own Hardware, System Software, Applications often to meet peak needs. Install, Configure, Test, Verify, Evaluate Manage . Finally, use it $.$(High CapEx,Cloud Computing Subscribe Use $ - pay

4、for what you use, based on QoS,Copyright 2012, Elsevier Inc. All rights reserved,Courtesy of Raj Buyya, 2012,整理課件,11,Basic Concept of Internet Clouds,整理課件,12,Transparent Cloud Computing Environment,整理課件,13,Cloud Computing Challenges: Dealing with too many issues (Courtesy of R. Buyya,整理課件,14,云計(jì)算技術(shù),三

5、大模式 四大部署 云計(jì)算的關(guān)鍵技術(shù),整理課件,15,云計(jì)算的三大交付模式,SaaS PaaS IaaS,整理課件,16,云計(jì)算的四大部署,公共云 私有云 社區(qū)云 混合云,整理課件,17,云計(jì)算的關(guān)鍵技術(shù),虛擬化技術(shù) 分布式計(jì)算和并行計(jì)算 分布式存儲 分布式海量數(shù)據(jù)管理,整理課件,18,1.虛擬化技術(shù),服務(wù)器虛擬化 存儲虛擬化 應(yīng)用虛擬化 平臺虛擬化 桌面虛擬化,整理課件,19,二、高校云計(jì)算必要性,減少硬件浪費(fèi) 綠色節(jié)能 提高效率 方便管理 資源共享 降低管理成本,整理課件,20,業(yè)務(wù)系統(tǒng)增多，只能通過采購新的硬件設(shè)備進(jìn)行業(yè)務(wù)支撐，硬件設(shè)備一多，導(dǎo)致“機(jī)房空間擁堵、能耗較高、不夠環(huán)?！?！ 數(shù)據(jù)

6、中心雖擁有眾多PC服務(wù)器，但服務(wù)器的資源利用率一直較低，存在較為嚴(yán)重的資源浪費(fèi)！ 數(shù)據(jù)中心軟硬件較多，但安全性重視程度不夠，缺少系統(tǒng)必備的可靠性設(shè)計(jì)，且設(shè)備宕機(jī)的風(fēng)險(xiǎn)較大！ 機(jī)房環(huán)境復(fù)雜，軟硬件設(shè)備繁多，系統(tǒng)管理人員難以有足夠的精力來進(jìn)行管理和維護(hù)！,整理課件,21,傳統(tǒng)數(shù)據(jù)中心現(xiàn)狀,服務(wù)器的百分比,主時(shí)間,高峰時(shí)間,20,40,60,80,0,0-10,20-30,40-50,60-70,80-90,CPU 利用率,當(dāng)前服務(wù)器利用率很低,為未使用的計(jì)算資源買單,x86服務(wù)器采購量/年,需要管理的服務(wù)器數(shù)量不斷增長,服務(wù)器部署：20,雜項(xiàng) 30,安全：10,支持：20,備份和監(jiān)視：15,在低價(jià)

7、值的任務(wù)上花費(fèi)了太多的時(shí)間，例如采購和調(diào)配 數(shù)據(jù)中心管理效率低,系統(tǒng)管理員如何使用他們的時(shí)間,低價(jià)值任務(wù),硬件采購：5,除了硬件的成本，服務(wù)器對空間的不斷侵占還帶來了管理成本的增長,2000,2002,2004,2006,2008,費(fèi)用（以十億美元為單位,管理和管制成本,服務(wù)器費(fèi)用,50,100,150,0,200,管理成本占TCO的大部分,100,整理課件,22,9,90%的服務(wù)器,90%的運(yùn)行時(shí)間,CPU占用率低于10,9,1,9-9-1原則,整理課件,23,1-主機(jī)存儲層面,大學(xué)數(shù)據(jù)中心現(xiàn)狀,整理課件,24,Datacenter and Server Cost Distribution,

8、整理課件,25,三、高校云計(jì)算可行性,數(shù)據(jù)中心虛擬化 私有云 PaaS SaaS,整理課件,26,虛擬機(jī)與數(shù)據(jù)中心虛擬化,Server Consolidation in Virtualized Datacenter Virtual Storage Provisioning and Deprovisioning Cloud Operating Systems for Virtual Datacenters Trust Management in virtualized Datacenters,整理課件,27,傳統(tǒng)計(jì)算機(jī)與虛擬機(jī)的區(qū)別,Courtesy of VMWare, 2008,整理課件,2

9、8,何為虛擬化 ,整理課件,29,Courtesy of VMWare, 2008,整理課件,30,VMM 作用 (1,Courtesy of Mendel Rosenblum, 2006,整理課件,31,VMM 作用(2,Courtesy of Mendel Rosenblum, 2006,整理課件,32,VMM 作用(3,Courtesy of Mendel Rosenblum, 2006,整理課件,33,VMM 作用(4,Courtesy of Mendel Rosenblum, 2006,整理課件,34,主要 的VMM 和 Hypervisor 軟件包,整理課件,35,存儲虛擬化,整理

10、課件,36,整理課件,37,整理課件,38,I/O虛擬化,整理課件,39,CPU, 存儲器和I/O虛擬化的總結(jié) CPU virtualization demands hardware-assisted traps of sensitive instructions by the VMM Memory virtualization demands special hardware support (shadow page tables by VMWare or extended page table by Intel) to help translate virtual address into

11、 physical address and machine memory in two stages. I/O virtualization is the most difficult one to realize due to the complexity if I/O service routines and the emulation needed between the guest OS and host OS,整理課件,40,虛擬機(jī)的熱遷移,整理課件,41,整理課件,42,整理課件,43,Resource provisioning methods,Demand-driven reso

12、urce provisioning based on the current utilization level of the allocated resources.(60%) Event-driven resource provisioning based on a specific time event.(predicte) Popularity-driven resource provisioning by popularity demand.(Internet searches,整理課件,44,虛擬機(jī)可信管理,整理課件,45,虛擬化市場 2006 to 2011 增長情況,整理課件,

13、46,私有云操作系統(tǒng),整理課件,47,整理課件,48,March 5, 2012,Prof. Kai Hwang, USC,Cloud Roles in The Future Internet,Courtesy of G. Xie and Z. Li 2012,整理課件,49,Three Aspects in Hardware that can benefit from Cloud Computing,The illusion of infinite computing resources available on demand, thereby eliminating the need fo

14、r cloud users to plan far ahead for resource provisioning. The elimination of an up-front commitment by cloud users, thereby allowing companies to start small and increase the hardware resources when needed in the future. The ability to pay the costs of computing resources on a short-term basis as needed (e.g., processors by the hour and storage by the day) and release them after done and thereby rewarding resource conservation,Courtesy of M. Ambrust, et al 2009,整理課件,50,整理課件,51,Cost-Effectiveness in Cloud Computingvs. Datacent