[論文筆記] Energy-aware resource allocation heuristics for efficient management of data centers for Clou

Timespan: 2.5 – 2.14
Anton Beloglazov, Jemal H. Abawajy, Rajkumar Buyya: Energy-aware resource allocation heuristics for efficient management of data centers for Cloud computing. Future Generation Comp. Syst. 28(5): 755-768 (2012) (gs:35)

    作者Anton Beloglazov是墨爾本大學的博士生，師從Rajkumar Buyya，同時還在IBM實習，研究興趣有：分布式系統、虛擬化、資料中心節能等。博士論文題目是：Energy and Performance Efficient Dynamic Consolidation of Virtual Machines in Cloud Data Centers，關注通過動態合并資料中心的虛擬機器來提高實體資源的利用率，在滿足QoS約束前提下降低能耗。目前，他還在參與OpenStack Neat（基於OpenStack的VM動態合并架構）。

    以下是他發表的論文：

Publication

Anton Beloglazov and Rajkumar Buyya, "Managing Overloaded Hosts for Dynamic Consolidation of Virtual Machines in Cloud Data Centers Under Quality of Service Constraints", IEEE Transactions on Parallel and Distributed Systems (TPDS), IEEE CS Press, USA, 2012Anton Beloglazov and Rajkumar Buyya, "OpenStack Neat: A Framework for Dynamic Consolidation of Virtual Machines in OpenStack Clouds - A Blueprint", Technical Report CLOUDS-TR-2012-4, Cloud Computing and Distributed Systems Laboratory, The University of Melbourne, August 14, 2012Anton Beloglazov, Sareh Fotuhi Piraghaj, Mohammed Alrokayan, and Rajkumar Buyya, "Deploying OpenStack on CentOS Using the KVM Hypervisor and GlusterFS Distributed File System", Technical Report CLOUDS-TR-2012-3, Cloud Computing and Distributed Systems Laboratory, The University of Melbourne, August 14, 2012Anton Beloglazov and Rajkumar Buyya, "Optimal Online Determin istic Algorithms and Adaptive Heuristics for Energy and Performance Efficient Dynamic Consolidation of Virtual Machines in Cloud Data Centers", Concurrency and Computation: Practice and Experience (CCPE), Volume 24, Issue 13, Pages: 1397-1420, John Wiley & Sons, Ltd, New York, USA, 2012Anton Beloglazov, Jemal Abawajy, and Rajkumar Buyya, "Energy-Aware Resource Allocation Heuristics for Efficient Management of Data Centers for Cloud Computing", The International Journal of Grid Computing and eScience, Future Generation Computer Systems (FGCS), Volume 28, Issue 5, Pages: 755-768, Elsevier Science, Amsterdam, The Netherlands, May 2012Kyong Hoon Kim, Anton Beloglazov and Rajkumar Buyya, "Power-Aware Provisioning of Virtual Machines for Real-time Cloud Services", Concurrency and Computation: Practice and Experience (CCPE), Volume 23, Number 13, Pages: 1492-1505, John Wiley & Sons, Ltd, New York, USA, 2011Anton Beloglazov and Rajkumar Buyya, "Energy-Efficient Consolidation of Virtual Machines in Cloud Data Centers", Proceedings of the IBM Collaborative Academia Research Exchange Workshop (I-CARE 2010), Bangalore, India, October 22, 2010Anton Beloglazov and Rajkumar Buyya, "Adaptive Threshold-Based Approach for Energy-Efficient Consolidation of Virtual Machines in Cloud Data Centers", Proceedings of the 8th International Workshop on Middleware for Grids, Clouds and e-Science (MGC 2010), Bangalore, India: ACM, November 29 — December 3, 2010Anton Beloglazov, Rajkumar Buyya, Young Choon Lee, and Albert Zomaya, "A Taxonomy and Survey of Energy-Efficient Data Centers and Cloud Computing Systems", Advances in Computers, Marvin V. Zelkowitz (editor), Volume 82, Pages: 47-111, ISSN: 0065-2458, Elsevier, 2011Rajkumar Buyya, Anton Beloglazov, and Jemal Abawajy, "Energy-Efficient Management of Data Center Resources for Cloud Computing: A Vision, Architectural Elements, and Open Challenges", Proceedings of the 2010 International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA 2010), Las Vegas, USA, July 12-15, 2010 — Keynote PaperAnton Beloglazov and Rajkumar Buyya, "Energy Efficient Allocation of Virtual Machines in Cloud Data Centers", In Proceedings of the 10th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid 2010), Melbourne, Australia, May 17-20, 2010Anton Beloglazov and Rajkumar Buyya, "Energy Efficient Resource Management in Virtualized Cloud Data Centers", IEEE TCSC Doctoral Symposium, In Proceedings of the 10th IEEE/ACM International Symposium on Cluster, Cloud, and Grid Computing (CCGrid 2010), Melbourne, Australia, May 17-20, 2010Rodrigo N. Calheiros, Rajiv Ranjan, Anton Beloglazov, Cesar A. F. De Rose, and Rajkumar Buyya, "CloudSim: A Toolkit for Modeling and Simulation of Cloud Computing Environments and Evaluation of Resource Provisioning Algorithms", Software: Practice and Experience (SPE), Volume 41, Number 1, Pages: 23-50, ISSN: 0038-0644, John Wiley & Sons, Ltd, New York, USA, January, 2011. DOI: 10.1002/spe.995Anton Beloglazov and Rajkumar Buyya, "Power and Performance Efficient Resource Management in Cloud Computing", in Proceedings of the IEEE Science and Engineering Graduate Research Expo 2009, Melbourne, Australia, 2009, pp. 38-40

以下是論文筆記：

1. 本文首先提出了一個節能雲端運算(energy-efficient Cloud computing)的架構架構(architectural framework)和原則。

系統架構圖(S3.1)

如所示，分為四層，其中Green Service Allocator一層角色眾多，承擔了對服務使用者請求進行分析、SLA協商、服務定價、VM調度、VM管理等功能。

2. (S3.2)中則提出了Power model，建立了能耗與CPU利用率的管理：

其中Pmax是伺服器完全利用時最高的能耗，k是空閑伺服器效率的能源比例（通常為70%)，u則是CPU的利用率。

(能耗跟記憶體、磁碟、網路使用方式也有關係，但主要與CPU有關，所以上面的公式裡只體現了CPU利用率，這個論斷見S3.2)

3. 在此基礎上，(S4)提出了資源供給與分配的演算法，以改進雲端運算環境的節能。該演算法是啟發學習法的(heuristics)，可以在確保滿足客戶QoS的前提下，使得資料中心的節能效果得到改進。

(S4.1) “VM placement”：是關於針對建立新的VM請求，分配到哪台PM使得能耗增加最小的問題。此問題被建模為“bin packing problem with variable bin sizes and prices”，使用了修改版的“Best Fit Decreasing (BFD)”演算法。

(S4.2)“VM selection”: 是關於如何最佳化當前的VM分配，以最佳化能耗的問題。主要分成兩個步驟：首先選擇遷移對象（一組VMs）；然後使用MBFD演算法，確定選出的遷移對象將被放置到哪些PMs上。

4. 針對“何時選擇哪些遷移對象的問題”，(S4.2)中提出了三種選擇策略（基本思想差不多）：

• the minimization of migrations(MM) policy
• the highest potential growth(HPG) policy
• the random choice(RC) policy

以MM策略為例，是該策略的規則：

大意如下：

    如果某一台PM的CPU利用率過高超出了上限，則找出一組個數最少的VM，這組VM就是遷移對象；

    如果某一台PM的CPU利用率過低達不到下限，則該PM上所有VM都是遷移對象。

5. (S5)進行了實驗驗證

(S5.1)中介紹了效能度量指標：

• total energy consumption
• SLA violation percentage
• number of VM migrations initiated by the VM manager
• average SLA violation

實驗是在類比平台(CloudSim toolkit)上進行的。具體的實驗過程和結果詳見原文。

6. 作者在(S6)提出了相關的open challenges

• optimization of VM placement according to the utilization of multiple system resources
• optimization of virtual network topologies
• optimization of thermal states and cooling system operation
• efficient consolidation of VMs for managing heterogeneous workloads
• a holistic approach to energy-aware resource management