近年來，資料倉儲技術漸趨成熟，許多行業均陸續完成系統導入應用，藉此改善或增進企業經營效率。在半導體製造產業中，也陸續興起導入資料倉儲系統的熱潮。然而，資料倉儲往往是巨額的資訊技術投資，要在維持晶圓廠持續生產運轉的條件下完成資料倉儲系統的導入，其挑戰更勝於全新的系統導入過程。 個案公司在以不影響生產活動，且維持晶圓廠每月滿載產能運轉的情況下導入資料倉儲系統，來滿足生產效率的提升需求。因此，系統導入方法的選定是相當重要的。本研究透過成功導入資料倉儲的案例探討與分析，並從中獲得理論與實務之交互印證，進一步分析出適配模式來制定資料倉儲建置計劃，其中，以資料萃取轉置處理、超資料管理機制之建構應用為主要改善重點。再配合晶圓廠生產管理作業需求，根據使用者需求的急迫程度，利用循環模式規劃出建置流程階段，逐步提升資料倉儲的資料覆蓋率來滿足使用者需求。 本研究所完成的具體成果，是以資料倉儲為核心基礎，建構雙軌制後端資料處理流程，包含資料模型設計以及資料萃取、轉換與載入等，並根據彙整之使用者應用需求，著手設計整合性資料分析平台。同時，透過超資料管理機制的規劃，具體提升資訊透明度，來優化資料倉儲的應用價值。在前端使用者介面的設計部份，則著墨在如何應用線上分析處理技術，提出資訊如何呈現的建議。整體導入計劃中的細部流程，以需求分析為始，完成整體系統分析，提出系統建構發展策略後，設計出資料模型與實體系統架構，包含資料萃取轉置機制與超資料管理、以及前端使用者介面設計評估，最後完成資料倉儲系統發展藍圖與導入階段的制定。本研究所完成之導入規劃將作為個案公司實務建置之參考依據。According to improve the performance of business operation and administration, more and more companies try to use system solutions to explain overall operation status information. With the enhancement in technique and theory, Data Warehouse becomes a very key player in this upcoming market. This upsurge is also impact the FAB manufacture business. However, the resource to implement Data Warehouse is truly big investment. And all of implementation step need to be completed within a factory running by full capacity. The challenge is more difficult to complete a brand new installation. In this case, the capacity of daily operation is not allow to be impacted. It is very impartment and difficult to figure out a methodology to execute the implementation without interfere the operation and also fulfill the desire of efficiency improvement. We try to approach from understand the daily operation and management process inside FAB and also study some success stories in other firm. We try to construct a best-fit solution that could be cross-verified by theory and real practices. The circulation pattern is used to build up the implementation flow. According to the priority of user’s requirements, we could set up a phase plan to implement. Through proceed key data scope one by one; we could increase the data coverage ratio to fulfill user’s need. The detail implementation steps are designed by through user request interview, system analysis and solution evaluating. The data model architecture and system architecture could be designed by bringing all of these outputs together. This architecture is also including data transformation management, metadata management and front-end user interface design. Then, we have a fully completed plan for data warehouse implementation. In this research, we build up an integrated analysis platform to support end users retrieve information on daily job. The core of this platform is based on a clear and well-defined data warehouse. The process includes data definition and retrieving, data transferring and transformation and data model define and design. At the same time, we also provide a mechanism to manage meta-data and provide a clearer view of information. All of those processes focus on nothing but one thing, to increase the add-on value of data ware house. The technique of on-line web base analysis behavior support is the key point of front-end design. It will be a great help to present information to be an accuracy advisement by using a user friendly operating flow and view.
