低碳雲端運算中心選址之研究
以台灣地區為例

雲端運算中心的選址是營運商的重大決策之一，在天災頻繁的台灣地區，必須考慮地震和氣候變遷引發之物理性風險，及是否能使用再生能源或自然資源做為降低日後營運成本的方式。本研究將以台灣地區為例，尋找通用之低碳雲端運算中心選址架構。研究準備先廣泛蒐集國內外相關文獻及標準規範，再遴選各相關領域具專業知識與豐富經驗的人員組成專家小組（expert panel），採用「修正型德菲法」（Modified Delphi method），進行第一階段問卷，並徵詢其意見直到各專家間的意見差異降至最低為止，協調一致性具體的共識後，彙整出低碳雲端運算中心選址評估架構表，共五大指標及三十二項評估準則。第二階段問卷主要受訪對象是對雲端運算中心設置位置有決定權或建議權的階層，此階段問卷採用李克特量表（Likert Scale），以低碳雲端運算中心選址評估架構表進行研究調查，結果呈現；因近年大型機房均面臨當地電力供應不足問題，雖然使用天然資源來降低雲端運算中心生命週期內的資源耗用與成本支出，但實務上因再生能源轉換效率不佳，導入新型技術對營運業主來說，會面臨初置成本較高的問題，此外引進自然資源做為降低碳排放的方式通常都有衍生技術問題需要解決，因此就重要程度而言，水電供應、交通便利、傳輸管道及後勤支援能力等基礎設施指標還是機房管理單位最重視的項目。本研究結果產出適合台灣地區通用的「低碳雲端運算中心選址架構」及低碳雲端運算中心選址指標權重表，並以各項指標權重平均值排序為；A:天然地理資源(23%)、B:基礎設施(28%)、鄰近人為風險(16%)、成本因素(20%)、地方政策及人力資源(13%)，並建議後續研究，進一步地理清驗證，本研究調查結果，期望在減少溫室氣體排放、降低持續營運風險及成本控管間取得平衡，並提供產業決策之參考依據。

關鍵字

選址；修正型德菲法；李克特量表；雲端運算；溫室氣體排放

並列摘要

The site selection of the Cloud Computing Center is one of the most challenging tasks with which decision maker for operation must face. Taiwan is located in a geographical region where has frequent natural disasters such as earthquake, typhoon, flooding, and so forth. Moreover, these physical risks are aggravated by global warming and climate change. Therefore, they need to be dealt with the consideration of using renewable energy or other natural resources for mitigating greenhouse gas emissions. The purpose of this study was to construct a generic framework for selecting a low-carbon cloud computing center site. The modified Delphi method was used to inquire domain experts for providing opinions on the critical factors/indicator of selecting low carbon cloud computing Center site in Taiwan. After the 1st phase of anonymous questionnaire survey and follow-up interview, a consensus was reached. A site selection evaluation table that includes five major indicators with 32 detail sub-criteria was generated. The 5 indicators are: natural geographical resources, Infrastructure, neighboring risk, cost factor, local policy and human resources. By using the evaluation table as a preliminary framework, the second phase of general questionnaire survey using Likert Scale method was conducted. Thirty-one experts who have the authority of making decision or suggestion on selecting the site of the cloud computing center were inquired. It is well known that large-scale computer centers are facing the problem of insufficient local power supply at Taiwan in recent years. Considering the use of the renewable energy to reduce carbon emission, other natural resource consumption and life cycle cost of the cloud computing center as well as knowing the present low efficiency of renewable energy technologies, the high initial costs and other unknown difficulties may encounter, the 2nd phase of questionnaire survey resulted in confirming the preliminary results of the 1st phase of Delphi survey and constructed a general-purpose of "low carbon cloud computing center site selection framework" along with an index weighting table. Based on the average value of weight ratios, the major indicators were: A, natural geographical resources (23%); B, Infrastructure (28 %); C, neighboring risk (16%); D, cost factor (20%); E, local policy and human resources (13%).