臺灣博碩士論文加值系統

隨著大數據時代的來臨，雲端運算顯得越來越重要，而用於雲端運算中的架構以MapReduce最為有名，例如Hadoop為目前最受廣泛使用且實現MapReduce架構之開源平台，而因為MapReduce為主從式架構，容易因為single point of failure單點失效，而造成系統問題，所以採用了peer-servicing的架構來避免因單點失效而造成問題，而因為是peer servicing，所以每個節點先計算分派到的工作，然後才形成最後結果，就像是霧一樣集結成雲，因此稱作霧運算。即每個使用者上傳工作後，會有peer node進行運算，而在藉由基礎設施即服務這類型的分布式霧運算技術，會有資源分配問題，所以對於這部分，是需考量進去的。
而在考慮服務導向的準則時，便採取Multi-Criteria Decision Making的想法來分配分布式霧運算資源，因Peer Node在分布式霧運算資源中也有限，所以採取MCDM來讓資源分配化，然後於P2P-MapReduce環境下進行運算工作。

Big data processing becomes more and more important, so the cloud computing is gaining increasing interest both in science and industry. One of the most popular framework in cloud computing is the MapReduce framework which exploits the advantage of distributed computing and highly reliability and scalability.
But MapReduce framework is the master-slave architecture, so if the master node failed, the system would occur error at unpredictable result. So using the peer-servicing architecture is the solution to avoid a single point of failure, we call it vapor computing. It means that peer nodes compute each task assigned to them until the whole job is completed. Consider for the situation that every peer node does the job until the result is formed just like that the vapor gather to transform into cloud. When every user submits the job, the peer nodes would contribute the computing resources for the incentives. But there will be a vapor computing resource allocation problem, so when considering service-oriented criteria, we will take the Multi-Criteria Decision Making method to allocate resources.

摘 要 I
Abstract II
ACKNOWLEDGEMENT III
TABLE OF CONTENTS IV
LIST OF TABLES VI
LIST OF FIGURES VII
Chapter 1. Introduction and Motivation 1
1.1. Introduction 1
1.2. Motivation 3
Chapter 2. Background and Related Work 5
2.1. MapReduce 5
2.1.1. Data flow 5
2.1.2. Map and Reduce phase 6
2.2. Hadoop 7
2.3. Resource Allocation 9
2.4. SPN-MR 10
2.5. Goal Programming Approach 11
2.5.1. Objective Function Reform 13
2.5.2. Linear Piecewise Approximation 13
2.5.3. Simplex Method Loop 14
2.5.4. Simplex Algorithm 15
2.6. P2P-MapReduce 17
2.7. Related Work 18
Chapter 3. System Design 21
3.1. Problem Description and Scenario 21
3.2. Parameter and Criteria Formulation 22
3.2.1. Parameter Definition 22
3.2.2. Modeling MapReduce to SPN-MR 23
3.2.3. Service-oriented Criteria 25
3.2.4. Infrastructure-oriented Criteria 27
3.3. Problem Formulation ＆ Goal Programming 28
3.4. Economic Activity 29
3.5. Flow Chart 30
Chapter 4. Performance Evaluation 33
4.1. Experiment environment 33
4.1.1. Benchmark 34
4.1.2. Executing Benchmark Execution Time 34
4.1.3. MCDM Setting 36
4.2. Results ＆ Evaluation 37
4.2.1. Utilization 37
4.2.2. Response Time 38
4.2.3. Cost Performance 40
Chapter 5. Conclusions and Future Work 43
References 44

[1]H. Manh Tran, S. Viet Uyen Ha, T. Kha Huynh, S. Thanh Le, “A feasible MapReduce peer-to-peer framework for distributed computing applications, Vietnam Journal of Computer Science, vol. 2, Issue 1 , pp. 57-66, Feb 2015.
[2]F. Marozzo, D. Talia, P. Trunfio, “P2P-MapReduce: Parallel data processing in dynamic Cloud environments, Journal of Computer and System Sciences, vol. 78, Issue 5, pp. 1382–1402, Sep. 2012
[3]F. Marozzo, D. Talia, P. Trunfio, “A Framework for Managing MapReduce Applications in Dynamic Distributed Environments, In proceeding of 19th International Euromicro Conference on Parallel, Distributed and Network-Based Processing, pp. 149-158, Feb. 2011.
[4]P. Patel, A. Ranabahu, and A. Sheth, “Service Level Agreement in Cloud Computing, In Proceedings of the Workshop on Best Practices in Cloud Computing: Implementation and Operational Implications for the Cloud at ACM International Conference on Object- Oriented Programming, Systems, Languages, and Applications, 2009
[5]Chih-Lun Chou, Gwo-Jiun Horng, Chieh-Ling Huang, and Wei-Chun Hsieh “Multi-Criteria Resource Brokering in Cloud Computing for Streaming Service, Journal of Mathematical Problems in Engineering, Article ID 823609, pp. 1-15, Mar.2015
[6]David Zimbeck, “Two Party double deposit trustless escrow in cryptographic networks and Bitcoin, http://www.BitHalo.org
[7]“Hadoop, http://hadoop.apache.org/
[8]J. Gray and A. Reuter, “Transaction Processing: Concepts and Techniques, 1993.
[9]C. Devlin, “SaaS Capacity Planning: Transaction Cost Analysis Revisited, http://msdn.microsoft.com/en-us/library/cc261632.aspx, 2008.
[10]H. Al-Hilali, D. Guimbellot, and M. Oslake Capacity Model for Internet Transactions, http://research.microsoft.com/pubs/69700/tr-99-18.doc, 1999.
[11]“Google MapReduce, http://research.google.com/archive/mapreduce.html
[12]Hsi-Chuan Wang, “Using Petri Net to Estimate Job Execution Time in MapReduce Model, Institute of Computer Science and Information Engineering, NCKU, 2013
[13]J. P. Ignizio, “A Review of Goal Programming: A Tool for Multiobjective Analysis, Journal of the Operational Research Society, vol. 29, 1978
[14]D.F. Jones and M. Tamiz, “Expanding the Flexibility of Goal Programming via Preference Modelling Techniques, Omega, vol. 23, 1995
[15]H. P. Williams, “Model Building in Mathematical Programming, Wiley, 1978
[16]W.L. Winston, “Operations Research: Applications and Algorithms, Duxbury Press, 3rd ed., 1997
[17]Y. Zhang, G. Huang, X. Liu, and H. Mei, “Integrating Resource Consumption and Allocation for Infrastructure Resources on-Demand, In Proceedings of the 3rd IEEE International Conference on Cloud Computing, 2010
[18]H. N. Van, F. D. Tran, and J. M. Menaud, “Performance and Power Management for Cloud Infrastructures, In Proceedings of the 3rd IEEE International Conference on Cloud Computing, 2010.
[19]“Management of Service Level Agreements for Multimedia Internet Service Using a Utility Model, IEEE Communications Magazine, vol. 39, no. 5, 2001.
[20]Chin-Fa Su, “ATGMR: An Adaptive Task Granularity Scheme for GPU-CPU MapReduce Clusters, Institute of Computer Science and Information Engineering, NCKU, 2014