VIETNAM NATIONAL UNIVERSITY, HANOI VIETNAM JAPAN UNIVERSITY PHAM TRUC QUYNH MULTICRITERIA ANALYSIS FOR HYPERSCALE DATA CENTERS PLACEMENT IN VIETNAM MASTER’S THESIS VIETNAM NATIONAL UNIVERSITY, HANOI VIETNAM JAPAN UNIVERSITY PHAM TRUC QUYNH MULTICRITERIA ANALYSIS FOR HYPERSCALE DATA CENTERS PLACEMENT IN VIETNAM MAJOR: BUSINESS ADMINISTRATION CODE: 8340101.01 RESEARCH SUPERVISORS: PROF. TANABU MONOTARI DR. LUU QUOC DAT Hanoi, 2022 STATEMENT OF COMMITMENT I have read and understood the plagiarism violations. I pledge with personal honor that this research result is my own and does not violate the Regulation on prevention of plagiarism in academic and scientific research activities at VNU Vietnam Japan University (Issued together with Decision No 700/QĐ-ĐHVN dated 30/09/2021 by the Rector of Vietnam Japan University).
Hanoi, 14h July 2022, Author Pham Truc Quynh i ACKNOWLEDGEMENT The completion of this master thesis is a long and enduring task which cannot happen without the support from various people. Firstly, my sincerest gratitude is owed to my supervisors, Professor Tanabu Monotari and Doctor Luu Quoc Dat, for their advice and guidance. Secondly, I am thankful to Professor Kurata Hisashi, Professor Suzuki Sadami and Professor Matsui Yoshiki for their advice and support in the Joint seminar for MIS- OM with Yokohama National University. Thirdly, I would like to extend my thanks to the Master of Business Administration Program, Vietnam Japan University staffs and Yokohama National University International/External Programs Office staffs, most especially Mrs.
Nguyen Thi Huong, the MBA program assistant for her constant guidance and encouragement. Finally, I would like to say thanks to my classmates, friends, and family for their continuing support for me in my studies. Hanoi, 14h July 2022, Author Pham Truc Quynh ii ABSTRACT Purpose – The research aims to address how the location problem for hyperscale data center can be efficiently solved in the current Vietnamese market. Design/methodology/approach – This study uses fuzzy analytic hierarchy process to analyze expert and non-experts’ opinions to calculate the criteria’s weights for the decision making model in selecting location for hyperscale data center.
Two approach for solving fuzziness are used and evaluated. The explored criteria are environment, accessibility to resources, cost, labor, and local government support. A simulation was created to demonstrate the process to evaluate locations for managers and other practioners for future use. Finding – The findings shows that even though the trends in the developed world is toward greener data center industry, in Vietnam, the most important criteria is still cost.
At the same time, the high skilled labor forces in the data center industry is still lacking. Practical implications – The final ranking of criteria create a roadmap for businesses to consider future actions in site selection for data center in Vietnamese market. Keywords: Data center, Hyperscale data center, location problem, MCDM, Multicriteria analysis, AHP, FAHP, Fuzzy AHP, Vietnam iii TABLE OF CONTENTS STATEMENT OF COMMITMENT. iii TABLE OF CONTENTS.
iv LIST OF TABLES. vi LIST OF FIGURES. ix LIST OF ABBREVIATIONS .2 Research problem formulation .5 Research paper structure. 4 CHAPTER 2: LITERATURE REVIEW.1 Facilities location problem .2 Data center location selection .3 Multicriteria decision making.4 Multicriteria decision making criteria for facilities location problem .2 Data collection for research necessity assessment .5 Data Collection for MCDM.
20 CHAPTER 4: ANALYSIS RESULTS AND DISCUSSION .1 Expert model analysis and discussion .1 Using Chang method .2 Using Hue et al method .2 Non-expert model analysis and discussion .3 Additional comments collected from experts .4 Validity and usefulness assessment interviews .5 Ranking of sub-criteria. 40 CHAPTER 5: CONCLUSION AND DICUSSION. 45 Appendix 1: QUESTIONNAIRE FOR EXPERTS. 53 Appendix 2: QUESTIONNAIRE FOR NON-EXPERTS.
65 Appendix 3: QUESTIONNAIRE FOR COMPARING LOCATION. 77 v LIST OF TABLES Table 2.1 Summary of site selection considerations from TIA 942 Standard Annex F .2 Selected literature review for criteria for facilities location problem .1 Saaty's preferences and triangular fuzzy conversion in the pair-wise comparison process.2 Reciprocal for Saaty's preferences and triangular fuzzy conversion in the pair- wise comparison process.4 Saaty's preferences and triangular fuzzy conversion in the pair-wise comparison process for non-expert model.1 Code for criteria and sub-criteria.2 Fuzzy comparison matrix and its priority vector for the first level’s criteria of expert model using Chang approach .3 Fuzzy comparison matrix and its priority vector for the second level’s criteria of expert model (Environment) using Chang approach .4 Fuzzy comparison matrix and its priority vector for the second level’s criteria of expert model (Accessibility to resources) using Chang approach .5 Fuzzy comparison matrix and its priority vector for the second level’s criteria of expert model (Cost) using Chang approach .6 Fuzzy comparison matrix and its priority vector for the second level’s criteria of expert model (Labor) using Chang approach .7 Fuzzy comparison matrix and its priority vector for the second level’s criteria of expert model (Local Government support) using Chang approach .8 Final weight for criteria and sub-criteria of expert model using Chang approach.9 Aggregated pair wise comparison matrices from experts for the first level’s criteria using Hue et al approach .10 Fuzzy comparison matrix and its priority vector for the first level’s criteria of expert model using Hue et al approach .11 Aggregated pair wise comparison matrices from experts for the second level’s criteria (Environment) using Hue et al approach .12 Fuzzy comparison matrix and its priority vector for the second level’s criteria of expert model (Environment) using Hue et al approach .13 Aggregated pair wise comparison matrices from experts for the second level’s criteria (Accessibility to resources) using Hue et al approach .14 Fuzzy comparison matrix and its priority vector for the second level’s criteria of expert model (Accessibility to resources) using Hue et al approach .15 Aggregated pair wise comparison matrices from experts for the second level’s criteria (Cost) using Hue et al approach .16 Fuzzy comparison matrix and its priority vector for the second level’s criteria of expert model (Cost) using Hue et al approach .17 Aggregated pair wise comparison matrices from experts for the second level’s criteria (Labor) using Hue et al approach .18 Fuzzy comparison matrix and its priority vector for the second level’s criteria of expert model (Labor) using Hue et al approach .19 Aggregated pair wise comparison matrices from experts for the second level’s criteria (Local Government support) using Hue et al approach .20 Fuzzy comparison matrix and its priority vector for the second level’s criteria of expert model (Local Government support) using Hue et al approach .21 Final weight for criteria and sub-criteria for expert model using Hue et al approach.22 Aggregated pair wise comparison matrices from non-experts for the first level’s criteria using Hue et al approach .23 Fuzzy comparison matrix and its priority vector for the first level’s criteria of non-expert model using Hue et al approach .24 Aggregated pair wise comparison matrices from non-experts for the second level’s criteria (Environment) using Hue et al approach .25 Fuzzy comparison matrix and its priority vector for the second level’s criteria of non-expert model (Environment) using Hue et al approach .26 Aggregated pair wise comparison matrices from non-experts for the second level’s criteria (Accessibility to resources) using Hue et al approach .27 Fuzzy comparison matrix and its priority vector for the second level’s criteria of non-expert model (Accessibility to resources) using Hue et al approach .28 Aggregated pair wise comparison matrices from non-experts for the second level’s criteria (Cost) using Hue et al approach .29 Fuzzy comparison matrix and its priority vector for the second level’s criteria of non-expert model (Cost) using Hue et al approach.30 Aggregated pair wise comparison matrices from non-experts for the second level’s criteria (Labor) using Hue et al approach .31 Fuzzy comparison matrix and its priority vector for the second level’s criteria of non-expert model (Labor) using Hue et al approach .32 Aggregated pair wise comparison matrices from non-experts for the second level’s criteria (Local Government support) using Hue et al approach .33 Fuzzy comparison matrix and its priority vector for the second level’s criteria of non-expert model (Local Government support) using Hue et al approach .34 Final weight for criteria and sub-criteria for non-expert model using Hue et al approach .35 Composite weights of all sub-criteria from expert model and their rankings against each other .36 Composite weights of expert model and their rankings for selecting current locations (Expert model minus Local government support sub-criteria) .37 Overall rating of 3 sites. 41 viii LIST OF FIGURES Figure 3.3 The comparison two fuzzy numbers .4 The distance between the centroid point Ci ( x S , y S ) and the minimum point i i G ( xmin , ymin ). 18 ix LIST OF ABBREVIATIONS MCDM Multi-criteria Decision Making AHP Analytic Hierarchy Process FAHP Fuzzy Analytic Hierarchy Process DC Data Center IDC Internet Data Center IT Information Technology ICT Information and Communication Technology AI Artificial Intelligent IoT Internet of Things M&E Mechanical and Electrical TIA Telecommunications Industry Association x CHAPTER 1.
INTRODUCTION This chapter provides a broad understanding of the data center industry in the world and in Vietnam as well as the challenge to select location for data centers. In addition, this chapter discusses the research objectives, and research scope.1 Research background Since the beginning of the 21st century, one of the biggest drivers of the economy worldwide is the IT industry. With more than seven billion devices connecting to the web, information is increasingly becoming a commodity that needs to be collected, stored, analyzed, and retrieved. The need for data management is increasingly important for businesses, with many choosing to outsource the task instead of keeping internal server systems.
According to Artizon’s “Data Center Market - Global Outlook & Forecast 2022-2027" report, the data center market is valued at 215 billion USD in 2021 and is forecasted to reach 288.3 billion USD in 2027. Also, according to the report, only in 2021, 400 facilities were built and growth are witnessed strongly in US, China, Japan, Australia, the UK, Germany, India, Saudi Arabia, South Africa, and Southeast Asian countries, namely, Indonesia, Malaysia, Philippines, Thailand, and Taiwan. Data center growth has been focused mostly in countries that give incentives either in tax exemption or in energy policies. Hyperscale data centers are data centers that are significantly larger than enterprise data centers with over 5,000 servers, and 10,000 square feet (Vertiv, 2021).
In recent years, the development of hyperscale data centers from Google, Facebook, AWS, Alibaba, and Microsoft are increasing. With that development, however, the data centers are facing a lot of challenges, especially regarding electricity usage. Data center is a facility that needs to be “on” almost 100% of the time, with Uptime Institute requiring 99.8 hours of downtime a year for the lowest tier, Tier 1. The highest tier, Tier 4, has expected uptime of 99.8 hours of downtime annually or 4 hour event in a 5 year period (Uptime Institute, 2009).
1% of global electricity is spent on data centers (Rooks, 2022). In the EU Commission study, data centers use 2.7% of the EU's total electricity in 2018 and will reach 3. Singapore uses 7% of its electricity for data center 1 in 2020 and employs a new requirement for new data center construction to be more energy sufficient in 2022 for newly opened data centers, removing the ban for new data centers in 2020 (Mah, 2022). To combat this challenge, many initiatives for greener data centers are being employed all over the world, such as The Science Based Targets initiative (SBTi), the Climate Neutral Data Center Pact, The Long Duration Energy Storage (LDES) Council, and the RE100.
The green data center market is valued at 35.58 billion USD in 2021 and is expected to reach 55.18 billion USD in 2027, according to Arizton’s “Green Data Center Market - Global Outlook & Forecast 2022-2027" report. Vietnam is a Lower Middle-Income Country, with aim to become a developed country in 2045. As such, Vietnamese Government is paying a lot of attention in the development of fintech, AI, E-Commerce, software outsourcing and education technology. With the Government push for 4.