衛(wèi) 星 張建軍 石 雷 翟 琰
?
云計(jì)算數(shù)據(jù)中心服務(wù)器數(shù)量動(dòng)態(tài)配置策略
衛(wèi) 星①②張建軍*①②石 雷①翟 琰①
①(合肥工業(yè)大學(xué)計(jì)算機(jī)與信息學(xué)院 合肥 230009)②(安全關(guān)鍵工業(yè)測(cè)控技術(shù)教育部工程研究中心 合肥 230009)
云計(jì)算數(shù)據(jù)中心由通過(guò)高速網(wǎng)絡(luò)連接的大量服務(wù)器構(gòu)成,一種有效的節(jié)能措施是維持與系統(tǒng)負(fù)載成比例的活躍服務(wù)器數(shù)量同時(shí)切換剩余服務(wù)器到空閑模式,由此分別產(chǎn)生操作能耗和切換能耗。該文研究如何動(dòng)態(tài)配置活躍服務(wù)器數(shù)量以最小化數(shù)據(jù)中心能耗(操作與切換能耗之和)的問(wèn)題。首先,建立了問(wèn)題的NP數(shù)學(xué)模型,并分析了無(wú)切換能耗情況下最優(yōu)解的特性;其次,通過(guò)消除整數(shù)動(dòng)態(tài)規(guī)劃的遞推過(guò)程,推導(dǎo)具有多項(xiàng)式復(fù)雜度的最優(yōu)靜態(tài)算法;最后,采用對(duì)未來(lái)負(fù)載的最壞預(yù)測(cè)結(jié)果作為約束制定了優(yōu)化在線策略。仿真結(jié)果表明,所提出的靜態(tài)最優(yōu)和動(dòng)態(tài)優(yōu)化策略能夠適應(yīng)外界負(fù)載的劇烈變化趨勢(shì)始終謹(jǐn)慎調(diào)整活躍服務(wù)器和休眠服務(wù)器的比例,以接近最優(yōu)的能耗代價(jià)維持?jǐn)?shù)據(jù)中心的平穩(wěn)運(yùn)行。
云計(jì)算;數(shù)據(jù)中心;活躍服務(wù)器;離線最優(yōu)算法;動(dòng)態(tài)規(guī)劃;在線算法
云計(jì)算通過(guò)整合存儲(chǔ)和計(jì)算能力有限的大量終端服務(wù)器,使得系統(tǒng)用戶(hù)只需通過(guò)網(wǎng)絡(luò)“透明”的訪問(wèn)其中一臺(tái)服務(wù)器就可獲得近乎無(wú)限的計(jì)算能力以及語(yǔ)音、視頻、信息搜索等服務(wù),而資源由云計(jì)算數(shù)據(jù)中心統(tǒng)一調(diào)度、組織和管理。Amazon, Google, IBM, Microsoft等相繼推出以集群計(jì)算為模型的云計(jì)算數(shù)據(jù)中心,采用層次結(jié)構(gòu)實(shí)現(xiàn)且承載的主要是客戶(hù)機(jī)/服務(wù)器模式應(yīng)用,具有如下典型特征:(1)數(shù)據(jù)中心內(nèi)部各服務(wù)器間具有高傳輸帶寬。(2)數(shù)據(jù)中心能夠?qū)崿F(xiàn)服務(wù)器和虛擬機(jī)的便捷配置和遷移。(3)數(shù)據(jù)中心支持?jǐn)?shù)十萬(wàn)甚至上百萬(wàn)臺(tái)的服務(wù)器,并允許增量的部署和擴(kuò)展,其服務(wù)能力遠(yuǎn)大于外部應(yīng)用需求。
本文研究如何動(dòng)態(tài)配置各時(shí)隙的活躍服務(wù)器數(shù)量從而最小化數(shù)據(jù)中心能耗的問(wèn)題。首先,從數(shù)據(jù)中心工作模式出發(fā),將任務(wù)分發(fā)策略簡(jiǎn)化為負(fù)載均衡方式并建立了問(wèn)題數(shù)學(xué)模型;其次,分析了無(wú)切換能耗情況下最優(yōu)解的特性,并給出了平周期與跟隨周期遞推法則;接下來(lái)通過(guò)消除整數(shù)動(dòng)態(tài)規(guī)劃的遞推過(guò)程,給出了具有多項(xiàng)式復(fù)雜度的靜態(tài)最優(yōu)算法;最后以未來(lái)負(fù)載的最壞預(yù)測(cè)結(jié)果為約束制定了在線算法。
從而操作能耗函數(shù)為
其次推導(dǎo)切換能耗函數(shù),由于活躍服務(wù)器切換到休眠模式需要負(fù)載遷移、機(jī)器折舊等損耗,而休眠模式到活躍模式的能耗極小可以忽略。切換能耗發(fā)生在相鄰時(shí)隙和之間,表達(dá)為,其中切換系數(shù)為正常數(shù)。由于“負(fù)載均衡”調(diào)度策略被廣泛接收是最優(yōu)分配方式[6,12],則任意服務(wù)器被分配到的負(fù)載為,且有。綜上所述,數(shù)據(jù)中心能耗最小化問(wèn)題可表述為
問(wèn)題1
3.1 無(wú)切換成本最優(yōu)解
問(wèn)題2
3.2 一般最優(yōu)解特性
圖1 最優(yōu)解特性—跟隨周期與平周期
問(wèn)題3
圖2 遞推用例
圖3 計(jì)算時(shí)存在的兩種情況
綜合以上兩種情況得
表1離線最優(yōu)算法偽代碼
表2在線優(yōu)化算法偽代碼
由于條件所限,仿真實(shí)驗(yàn)在 matlab 2013a環(huán)境下,采用離散事件動(dòng)態(tài)方法進(jìn)行仿真。整體運(yùn)行模式與流程類(lèi)似于數(shù)據(jù)中心的模型設(shè)定:服務(wù)器數(shù),服務(wù)容量,時(shí)隙總長(zhǎng);能耗參數(shù)分別設(shè)為,則操作能耗函數(shù)為。
6.1 離線仿真分析
圖4 不同負(fù)載變化情況下活躍服務(wù)器數(shù)
圖5 不同負(fù)載變化情況下系統(tǒng)最小能耗
6.2 在線仿真分析
圖6 在線算法與最優(yōu)離線算法的比較
圖7中online所示為100種場(chǎng)景下所得到的在線算法“性能比”曲線,可見(jiàn)其非常接近1。100種場(chǎng)景下的平均“性能比”為1.151,其中最大值和最小值分別為1.165和1.133。將作為一種在線算法進(jìn)行比較,所得到的“性能比”并不接近1,因?yàn)槠鋬H僅最小化了操作能耗。的“性能比”曲線其平均性能比為2.197,“性能比”最大值和最小值分別為2.297和2.111。而Lazy算法的“性能比”始終穩(wěn)定在1.45,比本文的在線優(yōu)化算法高出10%。由此可見(jiàn),同時(shí)考慮操作能耗和切換能耗是十分必要的,兩者必須同時(shí)達(dá)到均衡點(diǎn)才能使總體能耗最接近最優(yōu)離線算法所得到的最優(yōu)解。
圖7 100組工作負(fù)載場(chǎng)景下“性能比”曲線
本文研究如何靜態(tài)(離線)/動(dòng)態(tài)(在線)配置連續(xù)運(yùn)行時(shí)隙的活躍服務(wù)器數(shù)量,以最小化數(shù)據(jù)中心能耗的問(wèn)題。數(shù)值結(jié)果表明,本文所提出的離線最優(yōu)算法以較低的復(fù)雜度縮短了連續(xù)時(shí)隙運(yùn)行時(shí)延,同時(shí)符合活躍服務(wù)器數(shù)量需為整數(shù)的要求,為在線算法提供最優(yōu)參考依據(jù)。仿真分析表明,本文提出的在線優(yōu)化算法,能夠動(dòng)態(tài)適應(yīng)外界負(fù)載的劇烈變化趨勢(shì),始終較為謹(jǐn)慎地調(diào)整活躍服務(wù)器和休眠服務(wù)器的比例,始終以接近最優(yōu)的能耗代價(jià)維持?jǐn)?shù)據(jù)中心的平穩(wěn)運(yùn)行。進(jìn)一步的工作可以分為兩方面,一是以實(shí)際云計(jì)算數(shù)據(jù)中心的真實(shí)海量數(shù)據(jù)為來(lái)源,印證和提高算法的可行性與實(shí)用性,二是研究負(fù)載調(diào)度與活躍服務(wù)器配置聯(lián)合的綜合策略。
[1] Chong F T, Heck M J R, Ranganathan P,Data center energy efficiency: improving energy efficiency in data centers beyond technology scaling[J].&, 2014, 31(1): 93-104.
[2] Li Jian, Shuang Kai, Su Sen,Reducing operational costs through consolidation with resource prediction in the cloud[C]. 12th IEEE/ACM International Symposium on Cloud and Grid Computing (CCGrid), Ottawa, Canada, 2012: 793-798.
[3] Wang Lin, Zhang Fa, Arjona Aroca J,GreenDCN: a general framework for achieving energy efficiency in data center networks[J]., 2014, 32(1): 4-15.
[4] Urgaonkar R, Kozat U C, Igarashi K,Dynamic resource allocation and power management in virtualized data centers[C]. IEEE/IFIP Network Operations and Management Symposium (NOMS), Osaka, Japan, 2010: 479-486.
[5] Guenter B, Jain N, and Williams C. Managing cost, performance, and reliability tradeoffs for energy-aware server provisioning[C]. 2011 Proceedings of IEEEInternational Conference on Computer Communications (INFOCOM), Shanghai, China, 2011: 1332-1340.
[6] Qureshi A, Weber R, Balakrishnan H,Cutting the electric bill for internet-scale systems[J]., 2009, 39(4): 123-134.
[7] Guo Yuan-xiong and Fang Yu-guang. Electricity cost saving strategy in data centers by using energy storage[J]., 2013, 24(6): 1149-1160.
[8] Rao Lei, Liu Xue, Xie Le,Minimizing electricity cost: Optimization of distributed internet data centers in a multi-electricity market environment[C]. 2010 Proceedings of IEEE International Conference on Computer Communications (INFOCOM), San Diego, CA, USA, 2010: 1-9.
[9] Cao Jun-wei, Li Ke-qin and Stojmenovic I. Optimal power allocation and load distribution for multiple heterogeneous multi-core server processors across clouds and data centers[J]., 2014, 63(1): 45-58.
[10] Beloglazov A, Buyya R, Lee Y C,A taxonomy and survey of energy-efficient data centers and cloud computing systems[J]., 2011, 82(2): 47-111.
[11] Wang Kai, Lin Ming-hong, Ciucu F,Characterizing the impact of the workload on the value of dynamic resizing in data centers[C]. ACM SIGMETRICS/Performance, London, United Kingdom, 2012: 405-406.
[12] Rabbani M G, Zhani M F, and Boutaba R. On achieving high survivability in virtualized data centers[J]., 2014, E97B(1): 10-18.
[13] Liu Zhen-hua, Lin Ming-hong, Adam W,. Greening geographical load balancing[C]. Proceedings ACM SIGMETRICS, San Jose, CA, USA, 2011: 233-244.
[14] Mathew V, Sitaraman R K, and Shenoy P. Energy-aware load balancing in content delivery networks[C]. Proceedings of the ACM SIGMETRICS Joint International Conference on Measurement and Modeling of Computer Systems, Orlando, FL, USA, 2012: 954-962.
[15] Gandhi A, Gupta V, Harchol Balter M,Optimality analysis of energy-performance trade-off for server farm management[J]., 2010, 67(11): 1155-1171.
[16] Lin Ming-hong, Wierman A, Andrew L L H,. Dynamic right-sizing for power-proportional data centers[J]./, 2013, 21(5): 1378-1391.
[17] Michael R G and Johnson D S. Computers and Intractability: A Guide to the Theory of NP-completeness[M]. San Francisco: WH Freeman & Co., 1979: 206-218.
Dynamic Active Servers Allocating Policy for Cloud Computing Data Centers
Wei Xing①②Zhang Jian-jun①②Shi Lei①Zhai Yan①
①(,,230009,)②(-,230009,)
Cloud computing data centers generally consist of a large number of servers connected via high speed network. One promising approach to saving energy is to maintain enough active severs in proportion to system load, while switch left servers to idle mode whenever possible. Then operating cost and switching cost is brought about respectively. The problem of right-sizing active severs to minimize energy consumption (total cost of operating and switching) in data centers is discussed. Firstly, the NP-hard model is established, and the characteristics of the optimal solution when omitting the switching cost are analyzed. Then by revising the solution procedure carefully, the recursive procedure is successfully eliminated. The optimal static algorithm with polynomial complexity is achieved. Finally, the online strategy is developed using the worst predicting load as the constraints. Simulation results show that the proposed offline and online algorithm can adapt the dramatic trend of external load and always carefully adjust the proportion of active servers, to guarantee minimum power consumption with a smooth computing process.
Cloud computing; Data center; Active servers; Offline optimal algorithm; Dynamic programming; Online algorithm
TP393
A
1009-5896(2015)08-2007-07
10.11999/JEIT141286
張建軍 jianjun@hfut.edu.cn
2014-10-09收到,2015-04-16改回,2015-06-09網(wǎng)絡(luò)優(yōu)先出版
國(guó)家自然科學(xué)基金(61370088),國(guó)家國(guó)際科技合作專(zhuān)項(xiàng)項(xiàng)目(2014DFB10060)和中央高?;究蒲袠I(yè)務(wù)費(fèi)專(zhuān)項(xiàng)資金(2011HGBZ1321, 2012HGQC0012)資助課題
衛(wèi) 星: 男,1980年生,博士后,主要研究方向?yàn)橛?jì)算機(jī)網(wǎng)絡(luò)、離散事件動(dòng)態(tài)性能優(yōu)化.
張建軍: 男,1963年生,教授,主要研究方向?yàn)闄C(jī)電一體化、物聯(lián)網(wǎng)工程、新能源汽車(chē)、汽車(chē)電子.
石 雷: 男,1980年生,講師,主要研究方向?yàn)闊o(wú)線傳感網(wǎng).
翟 琰: 女,1977年生,講師,主要研究方向?yàn)槠?chē)電子、嵌入式系統(tǒng).